{ "cells": [ { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_DEVICE_ORDER\"] = \"PCI_BUS_ID\" \n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"0\"" ] }, { "cell_type": "code", "execution_count": 2, "metadata": { "scrolled": false }, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "Using TensorFlow backend.\n" ] } ], "source": [ "#Import all the necessary packages\n", "import pandas as pd\n", "import numpy as np\n", "import keras as k\n", "import tensorflow as tf\n", "from sklearn.preprocessing import MinMaxScaler\n", "import matplotlib\n", "import matplotlib.pyplot as plt\n", "from matplotlib import rcParams, cycler\n", "from sklearn.model_selection import train_test_split\n", "from tensorflow.keras.callbacks import EarlyStopping\n", "from tensorflow.keras.callbacks import ModelCheckpoint\n", "from tensorflow.keras.callbacks import ReduceLROnPlateau\n", "from tensorflow.keras.models import Model, load_model\n", "from tensorboard.plugins.hparams import api as hp\n", "from tensorflow.keras import regularizers\n", "from pickle import dump\n", "from sklearn.model_selection import KFold\n", "import tensorflowjs as tfjs\n", "import keract\n", "import datetime\n", "import math" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[name: \"/device:CPU:0\"\n", "device_type: \"CPU\"\n", "memory_limit: 268435456\n", "locality {\n", "}\n", "incarnation: 3595000761477948485\n", ", name: \"/device:GPU:0\"\n", "device_type: \"GPU\"\n", "memory_limit: 4937233203\n", "locality {\n", " bus_id: 1\n", " links {\n", " }\n", "}\n", "incarnation: 668913008403759742\n", "physical_device_desc: \"device: 0, name: GeForce GTX 1060 6GB, pci bus id: 0000:01:00.0, compute capability: 6.1\"\n", "]\n" ] } ], "source": [ "from tensorflow.python.client import device_lib \n", "print(device_lib.list_local_devices())" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "# Hyperparameters\n", "BATCH_SIZE = 128\n", "LEARNING_RATE = 0.001\n", "DROPOUT = 0.004\n", "VALIDATION_SPLIT = 0.1\n", "OPTIMIZER = 'adam'\n", "\n", "LAYER_1 = 900\n", "LAYER_2 = 150\n", "LAYER_3 = 700\n", "LAYER_4 = 550\n", "LAYER_5 = 950\n", "LAYER_6 = 950" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [], "source": [ "# Create a model\n", "def create_model():\n", " model = tf.keras.Sequential() \n", " model.add(tf.keras.layers.Dense(128, input_dim=len(keys) - 1, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(128, activation='relu'))\n", " model.add(tf.keras.layers.Dense(1))\n", "\n", " def rmsle(y_true, y_pred):\n", " y_pred_log = K.log(K.clip(y_pred, K.epsilon(), None) + 1.)\n", " y_true_log = K.log(K.clip(y_true, K.epsilon(), None) + 1.)\n", " return K.sqrt(K.mean(K.square(y_pred_log - y_true_log), axis = -1))\n", "\n", " model.compile(loss='mean_squared_error', \n", " optimizer=tf.keras.optimizers.Adam(lr=LEARNING_RATE, beta_1=0.9, beta_2=0.999, clipnorm=1.0)\n", " )\n", " \n", " return model" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Note: median values were scaled by multiplying by 0.0000011173 and adding -0.111732\n", "4043\n", "4043\n", "3226\n", "817\n" ] } ], "source": [ "# Load the dataset\n", "dataset = pd.read_csv(\"./Data/dataset_final.csv\")\n", "\n", "# Get keys\n", "keys = dataset.keys()\n", "\n", "# Target key\n", "target = 'deptFreePrice'\n", "\n", "# Initialize scaler\n", "scaler = MinMaxScaler(feature_range=(0, 1))\n", "\n", "# Scale both the training inputs and outputs\n", "scaled_train = scaler.fit_transform(dataset[keys])\n", "\n", "print(\"Note: median values were scaled by multiplying by {:.10f} and adding {:.6f}\".format(scaler.scale_[6], scaler.min_[6]))\n", "\n", "# Store the values to scale price back to understandable form\n", "multiplied_by = scaler.scale_[6]\n", "added = scaler.min_[6]\n", "\n", "scalerData = {'scale_': scaler.scale_, 'min_': scaler.min_}\n", "scalerDf = pd.DataFrame(scalerData, columns = ['scale_', 'min_'])\n", "\n", "\n", "scaled_train = pd.DataFrame(scaled_train, columns=keys)\n", "\n", "# Divide dataset into training and testing sets\n", "msk = np.random.rand(len(scaled_train)) < 0.8\n", "\n", "X, Y = scaled_train.drop(target, axis=1).values, scaled_train[target].values\n", "X_train_2 =scaled_train[msk].values\n", "\n", "X_train, X_test = scaled_train[msk].drop(target, axis=1).values, scaled_train[~msk].drop(target, axis=1).values\n", "Y_train, Y_test = scaled_train[msk][target].values, scaled_train[~msk][target].values\n", "\n", "print(len(X))\n", "print(len(Y))\n", "print(len(X_train))\n", "print(len(X_test))" ] }, { "cell_type": "code", "execution_count": 192, "metadata": { "scrolled": true }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Model: \"sequential_4\"\n", "_________________________________________________________________\n", "Layer (type) Output Shape Param # \n", "=================================================================\n", "dense_44 (Dense) (None, 128) 5504 \n", "_________________________________________________________________\n", "dense_45 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_46 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_47 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_48 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_49 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_50 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_51 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_52 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_53 (Dense) (None, 128) 16512 \n", "_________________________________________________________________\n", "dense_54 (Dense) (None, 1) 129 \n", "=================================================================\n", "Total params: 154,241\n", "Trainable params: 154,241\n", "Non-trainable params: 0\n", "_________________________________________________________________\n", "None\n", "Train on 2903 samples, validate on 323 samples\n", "Epoch 1/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0312 \n", "Epoch 00001: val_loss improved from inf to 0.02629, saving model to best_model.h5\n", "2903/2903 [==============================] - 1s 266us/sample - loss: 0.0282 - val_loss: 0.0263\n", "Epoch 2/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0116\n", "Epoch 00002: val_loss improved from 0.02629 to 0.01875, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 44us/sample - loss: 0.0111 - val_loss: 0.0188\n", "Epoch 3/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 0.0064\n", "Epoch 00003: val_loss improved from 0.01875 to 0.01634, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 43us/sample - loss: 0.0065 - val_loss: 0.0163\n", "Epoch 4/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0049\n", "Epoch 00004: val_loss improved from 0.01634 to 0.01453, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 42us/sample - loss: 0.0050 - val_loss: 0.0145\n", "Epoch 5/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0039\n", "Epoch 00005: val_loss improved from 0.01453 to 0.01223, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 40us/sample - loss: 0.0038 - val_loss: 0.0122\n", "Epoch 6/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0033\n", "Epoch 00006: val_loss improved from 0.01223 to 0.01057, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 42us/sample - loss: 0.0032 - val_loss: 0.0106\n", "Epoch 7/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 0.0032\n", "Epoch 00007: val_loss did not improve from 0.01057\n", "2903/2903 [==============================] - 0s 31us/sample - loss: 0.0034 - val_loss: 0.0114\n", "Epoch 8/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0031\n", "Epoch 00008: val_loss did not improve from 0.01057\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 0.0031 - val_loss: 0.0124\n", "Epoch 9/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0029\n", "Epoch 00009: val_loss did not improve from 0.01057\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 0.0030 - val_loss: 0.0118\n", "Epoch 10/2000\n", "1408/2903 [=============>................] - ETA: 0s - loss: 0.0024\n", "Epoch 00010: val_loss did not improve from 0.01057\n", "2903/2903 [==============================] - 0s 37us/sample - loss: 0.0027 - val_loss: 0.0113\n", "Epoch 11/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0029\n", "Epoch 00011: val_loss did not improve from 0.01057\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 0.0027 - val_loss: 0.0108\n", "Epoch 12/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 0.0022\n", "Epoch 00012: val_loss improved from 0.01057 to 0.01029, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 51us/sample - loss: 0.0024 - val_loss: 0.0103\n", "Epoch 13/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 0.0021\n", "Epoch 00013: val_loss did not improve from 0.01029\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 0.0022 - val_loss: 0.0146\n", "Epoch 14/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0023\n", "Epoch 00014: val_loss did not improve from 0.01029\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 0.0023 - val_loss: 0.0117\n", "Epoch 15/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0023\n", "Epoch 00015: val_loss did not improve from 0.01029\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 0.0023 - val_loss: 0.0107\n", "Epoch 16/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0019\n", "Epoch 00016: val_loss improved from 0.01029 to 0.00999, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 51us/sample - loss: 0.0022 - val_loss: 0.0100\n", "Epoch 17/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 0.0023\n", "Epoch 00017: val_loss did not improve from 0.00999\n", "2903/2903 [==============================] - 0s 37us/sample - loss: 0.0024 - val_loss: 0.0103\n", "Epoch 18/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 0.0021\n", "Epoch 00018: val_loss improved from 0.00999 to 0.00956, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 49us/sample - loss: 0.0021 - val_loss: 0.0096\n", "Epoch 19/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0022\n", "Epoch 00019: val_loss did not improve from 0.00956\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 0.0023 - val_loss: 0.0108\n", "Epoch 20/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0022\n", "Epoch 00020: val_loss did not improve from 0.00956\n", "2903/2903 [==============================] - 0s 30us/sample - loss: 0.0022 - val_loss: 0.0111\n", "Epoch 21/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0015\n", "Epoch 00021: val_loss improved from 0.00956 to 0.00909, saving model to best_model.h5\n", "2903/2903 [==============================] - 0s 46us/sample - loss: 0.0020 - val_loss: 0.0091\n", "Epoch 22/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0019\n", "Epoch 00022: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 0.0019 - val_loss: 0.0097\n", "Epoch 23/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0017\n", "Epoch 00023: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 0.0017 - val_loss: 0.0117\n", "Epoch 24/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 0.0021\n", "Epoch 00024: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 29us/sample - loss: 0.0021 - val_loss: 0.0097\n", "Epoch 25/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0017\n", "Epoch 00025: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 0.0016 - val_loss: 0.0095\n", "Epoch 26/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0015\n", "Epoch 00026: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 0.0015 - val_loss: 0.0114\n", "Epoch 27/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0015\n", "Epoch 00027: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 26us/sample - loss: 0.0015 - val_loss: 0.0117\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "Epoch 28/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0017\n", "Epoch 00028: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 0.0018 - val_loss: 0.0117\n", "Epoch 29/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 0.0016\n", "Epoch 00029: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 31us/sample - loss: 0.0017 - val_loss: 0.0105\n", "Epoch 30/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0015\n", "Epoch 00030: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 0.0016 - val_loss: 0.0103\n", "Epoch 31/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 0.0016\n", "Epoch 00031: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 36us/sample - loss: 0.0016 - val_loss: 0.0099\n", "Epoch 32/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0014\n", "Epoch 00032: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 0.0015 - val_loss: 0.0110\n", "Epoch 33/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 0.0017\n", "Epoch 00033: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 39us/sample - loss: 0.0016 - val_loss: 0.0106\n", "Epoch 34/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0013\n", "Epoch 00034: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 0.0013 - val_loss: 0.0114\n", "Epoch 35/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0012\n", "Epoch 00035: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 0.0013 - val_loss: 0.0102\n", "Epoch 36/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0013\n", "Epoch 00036: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 0.0013 - val_loss: 0.0096\n", "Epoch 37/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0012 \n", "Epoch 00037: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 0.0013 - val_loss: 0.0102\n", "Epoch 38/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0014 \n", "Epoch 00038: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 0.0014 - val_loss: 0.0104\n", "Epoch 39/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0015\n", "Epoch 00039: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 0.0015 - val_loss: 0.0110\n", "Epoch 40/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0014\n", "Epoch 00040: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 0.0014 - val_loss: 0.0109\n", "Epoch 41/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0013\n", "Epoch 00041: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 0.0013 - val_loss: 0.0105\n", "Epoch 42/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 0.0013\n", "Epoch 00042: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 30us/sample - loss: 0.0012 - val_loss: 0.0112\n", "Epoch 43/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0011 \n", "Epoch 00043: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 0.0011 - val_loss: 0.0113\n", "Epoch 44/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0013 \n", "Epoch 00044: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 0.0013 - val_loss: 0.0118\n", "Epoch 45/2000\n", "2560/2903 [=========================>....] - ETA: 0s - loss: 0.0013\n", "Epoch 00045: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 25us/sample - loss: 0.0013 - val_loss: 0.0114\n", "Epoch 46/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 0.0011 \n", "Epoch 00046: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 29us/sample - loss: 0.0011 - val_loss: 0.0113\n", "Epoch 47/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0011\n", "Epoch 00047: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 29us/sample - loss: 0.0011 - val_loss: 0.0109\n", "Epoch 48/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0011 \n", "Epoch 00048: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 26us/sample - loss: 0.0011 - val_loss: 0.0111\n", "Epoch 49/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 0.0011 \n", "Epoch 00049: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 0.0011 - val_loss: 0.0118\n", "Epoch 50/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0010\n", "Epoch 00050: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 0.0011 - val_loss: 0.0117\n", "Epoch 51/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 9.9550e-04\n", "Epoch 00051: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 29us/sample - loss: 0.0011 - val_loss: 0.0111\n", "Epoch 52/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 0.0011 \n", "Epoch 00052: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 0.0011 - val_loss: 0.0105\n", "Epoch 53/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 9.7608e-04\n", "Epoch 00053: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 9.7248e-04 - val_loss: 0.0121\n", "Epoch 54/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 9.1915e-04\n", "Epoch 00054: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 31us/sample - loss: 9.8284e-04 - val_loss: 0.0116\n", "Epoch 55/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0010\n", "Epoch 00055: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 0.0010 - val_loss: 0.0115\n", "Epoch 56/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 9.5331e-04\n", "Epoch 00056: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 9.7758e-04 - val_loss: 0.0111\n", "Epoch 57/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 9.8361e-04\n", "Epoch 00057: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 9.6959e-04 - val_loss: 0.0113\n", "Epoch 58/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 9.1950e-04\n", "Epoch 00058: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 9.5159e-04 - val_loss: 0.0105\n", "Epoch 59/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 8.8970e-04\n", "Epoch 00059: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 36us/sample - loss: 0.0010 - val_loss: 0.0116\n", "Epoch 60/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 8.1791e-04\n", "Epoch 00060: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 36us/sample - loss: 9.2781e-04 - val_loss: 0.0114\n", "Epoch 61/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 9.4657e-04\n", "Epoch 00061: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 38us/sample - loss: 9.3655e-04 - val_loss: 0.0119\n", "Epoch 62/2000\n", "1664/2903 [================>.............] - ETA: 0s - loss: 0.0011 \n", "Epoch 00062: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 39us/sample - loss: 0.0011 - val_loss: 0.0132\n", "Epoch 63/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 0.0013\n", "Epoch 00063: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 40us/sample - loss: 0.0013 - val_loss: 0.0110\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "Epoch 64/2000\n", "1280/2903 [============>.................] - ETA: 0s - loss: 8.7640e-04\n", "Epoch 00064: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 42us/sample - loss: 8.4999e-04 - val_loss: 0.0104\n", "Epoch 65/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 7.9611e-04\n", "Epoch 00065: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 8.7371e-04 - val_loss: 0.0108\n", "Epoch 66/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 8.4635e-04\n", "Epoch 00066: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 9.1774e-04 - val_loss: 0.0123\n", "Epoch 67/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 8.8140e-04\n", "Epoch 00067: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 8.7677e-04 - val_loss: 0.0115\n", "Epoch 68/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 8.1287e-04\n", "Epoch 00068: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 26us/sample - loss: 8.2816e-04 - val_loss: 0.0119\n", "Epoch 69/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 9.5661e-04\n", "Epoch 00069: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 9.4845e-04 - val_loss: 0.0111\n", "Epoch 70/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 9.4214e-04\n", "Epoch 00070: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 9.2418e-04 - val_loss: 0.0115\n", "Epoch 71/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 8.8652e-04\n", "Epoch 00071: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 30us/sample - loss: 9.3996e-04 - val_loss: 0.0120\n", "Epoch 72/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 0.0015\n", "Epoch 00072: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 0.0014 - val_loss: 0.0117\n", "Epoch 73/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 9.7117e-04\n", "Epoch 00073: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 9.7685e-04 - val_loss: 0.0109\n", "Epoch 74/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 8.1636e-04\n", "Epoch 00074: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 31us/sample - loss: 8.3132e-04 - val_loss: 0.0116\n", "Epoch 75/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 8.1337e-04\n", "Epoch 00075: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 9.9363e-04 - val_loss: 0.0125\n", "Epoch 76/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 0.0013\n", "Epoch 00076: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 0.0012 - val_loss: 0.0114\n", "Epoch 77/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 7.2394e-04\n", "Epoch 00077: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 30us/sample - loss: 7.4719e-04 - val_loss: 0.0113\n", "Epoch 78/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 7.8732e-04\n", "Epoch 00078: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 8.3259e-04 - val_loss: 0.0106\n", "Epoch 79/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 7.6195e-04\n", "Epoch 00079: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 8.6152e-04 - val_loss: 0.0128\n", "Epoch 80/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 8.4719e-04\n", "Epoch 00080: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 8.5586e-04 - val_loss: 0.0123\n", "Epoch 81/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 7.7491e-04\n", "Epoch 00081: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 31us/sample - loss: 8.3792e-04 - val_loss: 0.0107\n", "Epoch 82/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 7.1667e-04\n", "Epoch 00082: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 7.4843e-04 - val_loss: 0.0114\n", "Epoch 83/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 7.4752e-04\n", "Epoch 00083: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 7.1481e-04 - val_loss: 0.0117\n", "Epoch 84/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 6.5041e-04\n", "Epoch 00084: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 7.3454e-04 - val_loss: 0.0126\n", "Epoch 85/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 7.7535e-04\n", "Epoch 00085: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 36us/sample - loss: 7.4483e-04 - val_loss: 0.0114\n", "Epoch 86/2000\n", "1664/2903 [================>.............] - ETA: 0s - loss: 7.5964e-04\n", "Epoch 00086: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 37us/sample - loss: 7.8158e-04 - val_loss: 0.0122\n", "Epoch 87/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 7.6849e-04\n", "Epoch 00087: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 37us/sample - loss: 7.2185e-04 - val_loss: 0.0135\n", "Epoch 88/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 6.5576e-04\n", "Epoch 00088: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 36us/sample - loss: 6.4484e-04 - val_loss: 0.0107\n", "Epoch 89/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 6.7173e-04\n", "Epoch 00089: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 6.7766e-04 - val_loss: 0.0119\n", "Epoch 90/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 5.9617e-04\n", "Epoch 00090: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 29us/sample - loss: 6.3005e-04 - val_loss: 0.0116\n", "Epoch 91/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 7.2313e-04\n", "Epoch 00091: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 7.1930e-04 - val_loss: 0.0124\n", "Epoch 92/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 7.0330e-04\n", "Epoch 00092: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 7.2412e-04 - val_loss: 0.0112\n", "Epoch 93/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 6.7305e-04\n", "Epoch 00093: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 30us/sample - loss: 7.0767e-04 - val_loss: 0.0108\n", "Epoch 94/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 6.9092e-04\n", "Epoch 00094: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 29us/sample - loss: 6.9017e-04 - val_loss: 0.0115\n", "Epoch 95/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 6.5161e-04\n", "Epoch 00095: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 6.3312e-04 - val_loss: 0.0114\n", "Epoch 96/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 6.4123e-04\n", "Epoch 00096: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 29us/sample - loss: 6.2545e-04 - val_loss: 0.0115\n", "Epoch 97/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 5.6344e-04\n", "Epoch 00097: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 6.0680e-04 - val_loss: 0.0108\n", "Epoch 98/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 8.2574e-04\n", "Epoch 00098: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 26us/sample - loss: 8.1689e-04 - val_loss: 0.0110\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "Epoch 99/2000\n", "2560/2903 [=========================>....] - ETA: 0s - loss: 7.3464e-04\n", "Epoch 00099: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 25us/sample - loss: 7.2157e-04 - val_loss: 0.0132\n", "Epoch 100/2000\n", "2560/2903 [=========================>....] - ETA: 0s - loss: 7.1252e-04\n", "Epoch 00100: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 25us/sample - loss: 7.1788e-04 - val_loss: 0.0112\n", "Epoch 101/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 6.5079e-04\n", "Epoch 00101: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 6.2962e-04 - val_loss: 0.0123\n", "Epoch 102/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 6.0506e-04\n", "Epoch 00102: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 32us/sample - loss: 6.0545e-04 - val_loss: 0.0112\n", "Epoch 103/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 6.0205e-04\n", "Epoch 00103: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 6.2116e-04 - val_loss: 0.0119\n", "Epoch 104/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 6.0589e-04\n", "Epoch 00104: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 6.4509e-04 - val_loss: 0.0117\n", "Epoch 105/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 6.4410e-04\n", "Epoch 00105: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 6.4799e-04 - val_loss: 0.0111\n", "Epoch 106/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 6.1937e-04\n", "Epoch 00106: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 6.1239e-04 - val_loss: 0.0120\n", "Epoch 107/2000\n", "1664/2903 [================>.............] - ETA: 0s - loss: 6.6325e-04\n", "Epoch 00107: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 36us/sample - loss: 6.1345e-04 - val_loss: 0.0113\n", "Epoch 108/2000\n", "1792/2903 [=================>............] - ETA: 0s - loss: 5.1688e-04\n", "Epoch 00108: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 35us/sample - loss: 5.3364e-04 - val_loss: 0.0125\n", "Epoch 109/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 6.6596e-04\n", "Epoch 00109: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 6.7963e-04 - val_loss: 0.0115\n", "Epoch 110/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 6.4278e-04\n", "Epoch 00110: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 6.0299e-04 - val_loss: 0.0112\n", "Epoch 111/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 5.3297e-04\n", "Epoch 00111: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 5.5112e-04 - val_loss: 0.0115\n", "Epoch 112/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 5.8987e-04\n", "Epoch 00112: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 34us/sample - loss: 6.0769e-04 - val_loss: 0.0130\n", "Epoch 113/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 5.5111e-04\n", "Epoch 00113: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 5.6370e-04 - val_loss: 0.0121\n", "Epoch 114/2000\n", "1920/2903 [==================>...........] - ETA: 0s - loss: 4.8050e-04\n", "Epoch 00114: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 4.8669e-04 - val_loss: 0.0116\n", "Epoch 115/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 4.4327e-04\n", "Epoch 00115: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 30us/sample - loss: 4.9087e-04 - val_loss: 0.0120\n", "Epoch 116/2000\n", "2304/2903 [======================>.......] - ETA: 0s - loss: 6.1409e-04\n", "Epoch 00116: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 28us/sample - loss: 6.5732e-04 - val_loss: 0.0120\n", "Epoch 117/2000\n", "2176/2903 [=====================>........] - ETA: 0s - loss: 6.1762e-04\n", "Epoch 00117: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 31us/sample - loss: 5.7927e-04 - val_loss: 0.0121\n", "Epoch 118/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 5.1383e-04\n", "Epoch 00118: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 33us/sample - loss: 5.3133e-04 - val_loss: 0.0120\n", "Epoch 119/2000\n", "2048/2903 [====================>.........] - ETA: 0s - loss: 5.2612e-04\n", "Epoch 00119: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 30us/sample - loss: 5.3072e-04 - val_loss: 0.0115\n", "Epoch 120/2000\n", "2432/2903 [========================>.....] - ETA: 0s - loss: 4.7557e-04\n", "Epoch 00120: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 27us/sample - loss: 4.9869e-04 - val_loss: 0.0116\n", "Epoch 121/2000\n", "2560/2903 [=========================>....] - ETA: 0s - loss: 4.7564e-04\n", "Epoch 00121: val_loss did not improve from 0.00909\n", "2903/2903 [==============================] - 0s 26us/sample - loss: 4.7991e-04 - val_loss: 0.0120\n", "Epoch 00121: early stopping\n" ] } ], "source": [ "# Callbacks for the model\n", "es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=100)\n", "mc = ModelCheckpoint('best_model.h5', monitor='val_loss', mode='min', verbose=1, save_best_only=True)\n", "reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.90,\n", " patience=10, min_lr=0.000001)\n", "n_split = 10\n", "h = []\n", "e = []\n", "j = 0\n", "badIndexes = {}\n", "#for train_index,test_index in KFold(n_split, True, 1).split(X_train_2):\n", " #x_train,x_test=X[train_index],X[test_index]\n", " #y_train,y_test=Y[train_index],Y[test_index]\n", " \n", " #model=create_model()\n", " #h.append(model.fit(x_train, y_train, batch_size=16, verbose=1,epochs=1000,callbacks=[mc]))\n", " #e.append({'Fold': j, 'MSE': model.evaluate(x_test,y_test)})\n", " #j += 1\n", " #if(j==2):\n", " # badIndexes = {'x':train_index, 'y': test_index}\n", " #print('Model evaluation ',model.evaluate(x_test,y_test))\n", "\n", "#log_dir = os.path.join(\"./logs/scalars/\")\n", "#tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=log_dir, histogram_freq=1, profile_batch=0)\n", "\n", "# Create and run the model\n", "model = create_model()\n", "print(model.summary())\n", "h = model.fit(\n", " X_train,\n", " Y_train,\n", " validation_split=VALIDATION_SPLIT,\n", " batch_size=BATCH_SIZE,\n", " epochs=2000,\n", " shuffle=True,\n", " verbose=1,\n", " callbacks=[es, mc, reduce_lr \n", " #tensorboard_callback\n", " ]\n", ")\n", "\n", "# Save the best model for later use\n", "best_model_org = load_model('best_model.h5')\n" ] }, { "cell_type": "code", "execution_count": 196, "metadata": {}, "outputs": [], "source": [ "#Best model found by using WANDB developer tool\n", "best_model = load_model('./models/wandb/model-best.h5')\n", "best_model_2 = load_model('./models/wandb/model-best_2.h5')\n", "best_model_3 = load_model('./models/wandb/model-best_3.h5')\n", "best_model_4 = load_model('./models/wandb/model-best_4.h5')\n", "best_model_5 = load_model('./models/wandb/model-best_5.h5')" ] }, { "cell_type": "code", "execution_count": 197, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "MSE for the data set is: 0.001658357523372472\n" ] }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "#Print the training history\n", "fig, ax = plt.subplots(figsize=(20, 10))\n", "lines = ax.plot(h.history['loss'], label='Loss')\n", "lines = ax.plot(h.history['val_loss'], label='val_loss')\n", "ax.legend(loc='upper right',fontsize=15)\n", "\n", "test_error_rate = best_model.evaluate(X_train, Y_train, verbose=0)\n", "print(\"MSE for the data set is: {}\".format(test_error_rate))" ] }, { "cell_type": "code", "execution_count": 211, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "23320.873046875\n", "8.325565338134766\n", "33232.23302758935\n", "(817,)\n", "(817, 1)\n", "0.9483329544524007\n" ] } ], "source": [ "predictors = keys.drop(target)\n", "\n", "# Make predictions from test set\n", "prediction = best_model.predict(X_test)\n", "\n", "MAE = 0\n", "RME = 0\n", "RMSE = 0\n", "\n", "pred_p = []\n", "pred_p2 = []\n", "pred_p3 = []\n", "org_p = []\n", "\n", "# Scale target values back to normal\n", "for p in Y_test:\n", " y_2 = p\n", " y_2 -= added\n", " y_2 /= multiplied_by\n", " org_p.append(y_2)\n", " \n", "org_p = np.asarray(org_p)\n", "\n", "# Scale predicted values back to normal\n", "for p in prediction:\n", " y_2 = p\n", " y_2 -= added\n", " y_2 /= multiplied_by\n", " pred_p.append(y_2)\n", " \n", "# Calculate metrics for sensitivity analysis\n", "for i in range(len(pred_p)):\n", " MAE += abs(pred_p[i] - org_p[i])\n", " RME += abs(pred_p[i] - org_p[i]) / org_p[i] * 100\n", " RMSE += (pred_p[i] - org_p[i])**2\n", " \n", "\n", "MAE = float(MAE / len(pred_p))\n", "RME = float(RME / len(pred_p))\n", "RMSE = math.sqrt(RMSE / len(pred_p))\n", "print(MAE)\n", "print(RME)\n", "print(RMSE)\n", "newArray = np.asarray(pred_p)\n", "#print(\"\\nRMSPE for the data set is: {0:.2f}%\".format(error_p))\n", "\n", "print(org_p.shape)\n", "print(newArray.shape)\n", "\n", "# Calculate r squared\n", "org_p = org_p.reshape((817,))\n", "newArray = newArray.reshape((817,))\n", "correlation_matrix = np.corrcoef(org_p, newArray)\n", "correlation_xy = correlation_matrix[0,1]\n", "r_squared = correlation_xy**2\n", "\n", "print(r_squared)" ] }, { "cell_type": "code", "execution_count": 212, "metadata": { "scrolled": false }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "314070.84\n", "283000.0\n", "(817,)\n", "1\n", "Min value - 115826.23 Min original - 101000.0\n", "Max value - 960688.75 Max original - 990000.0\n", "Mean - 311228.25 Mean original - 305065.43772337824\n", "Std - 139710.78 Std original - 143663.35221658647\n", "DIF: -0.00027964749506399506\n", "\n", "MSE for the data set is: 0.0014\n" ] }, { "data": { "text/plain": [ "(array([ 0., 200000., 400000., 600000., 800000., 1000000.,\n", " 1200000.]), )" ] }, "execution_count": 212, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "from matplotlib import rcParams, cycler\n", "\n", "print(\"Min value -\", min(newArray), \" Min original - \", min(org_p))\n", "print(\"Max value -\", max(newArray), \" Max original - \", max(org_p))\n", "print(\"Mean -\", np.nanmean(newArray), \" Mean original - \", np.nanmean(org_p))\n", "print(\"Std -\", np.nanstd(newArray), \" Std original - \", np.nanstd(org_p))\n", "\n", "error_rate = best_model.evaluate(X_test, Y_test, verbose=0)\n", "print(\"DIF: \", (error_rate - test_error_rate))\n", "print(\"\\nMSE for the data set is: {0:.4f}\".format(error_rate))\n", "\n", "cmap = plt.cm.coolwarm\n", "rcParams['axes.prop_cycle'] = cycler(color=cmap(np.linspace(0, 1, 2)))\n", "\n", "# Print the predictions and targeted prices\n", "fig, ax = plt.subplots(figsize=(15, 15))\n", "lines = ax.plot(org_p, org_p, label='Predicted', linewidth=2)\n", "ax.scatter(org_p, newArray, color='r')\n", "ax.set_xlabel('Original price',fontsize=18)\n", "ax.set_ylabel('Predicted price',fontsize=18)\n", "\n", "txt = '\\n'.join((\"MAE: {0:.0f} €\".format(MAE), \"RME: {0:.2f}%\".format(RME), \"RMSE: {0:.0f} €\".format(RMSE), \"R²: {0:.2f}\".format(r_squared)))\n", "ax.text(0.01, 0.91, txt, fontsize=16,\n", " verticalalignment='bottom', transform=ax.transAxes)\n", "\n", "plt.xticks(fontsize=14)\n", "plt.yticks(fontsize=14)" ] }, { "cell_type": "code", "execution_count": 214, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[[18329.39915531517, 9.115760385440007, 24365.823791975272, 0.8800033819288181], [20606.306053314696, 8.154917373656778, 28222.32272844262, 0.9437376308301402], [25804.80205195783, 8.119345543883703, 38430.00922171203, 0.9349564534440795], [29743.508049242424, 9.045208528737414, 38703.161362382496, 0.9530557636322816], [24786.3566015625, 5.292388211234174, 35173.41986743031, 0.9525933284681801], [45571.90401785716, 14.961291596888406, 63119.48403356005, 0.8913788826248188]]\n" ] } ], "source": [ "# Calculate sensitivity analysis for each room type\n", "rooms = []\n", "pred_sep = [[], [], [], [], [], []]\n", "org_sep = [[], [], [], [], [], []]\n", "stats = []\n", "for i in range(len(X_test)):\n", " scaled = math.floor((X_test[i][3] - scaler.min_[3]) / scaler.scale_[3])\n", " rooms.append(scaled)\n", " \n", "for i in range(len(newArray)):\n", " if rooms[i] == 1:\n", " pred_sep[0].append(newArray[i])\n", " org_sep[0].append(org_p[i])\n", " elif rooms[i] == 2:\n", " pred_sep[1].append(newArray[i])\n", " org_sep[1].append(org_p[i])\n", " elif rooms[i] == 3:\n", " pred_sep[2].append(newArray[i])\n", " org_sep[2].append(org_p[i])\n", " elif rooms[i] == 4:\n", " pred_sep[3].append(newArray[i])\n", " org_sep[3].append(org_p[i])\n", " elif rooms[i] == 5:\n", " pred_sep[4].append(newArray[i])\n", " org_sep[4].append(org_p[i])\n", " elif rooms[i] == 6:\n", " pred_sep[5].append(newArray[i])\n", " org_sep[5].append(org_p[i])\n", " \n", "\n", "for n in range(len(pred_sep)):\n", " for i in range(len(pred_sep[n])):\n", " MAE += abs(pred_sep[n][i] - org_sep[n][i])\n", " RME += abs(pred_sep[n][i] / org_sep[n][i] - 1) * 100\n", " RMSE += (pred_sep[n][i] - org_sep[n][i])**2\n", "\n", " correlation_matrix = np.corrcoef(org_sep[n], pred_sep[n])\n", " correlation_xy = correlation_matrix[0,1]\n", " r_squared = correlation_xy**2\n", " MAE = MAE / len(pred_sep[n])\n", " RME = RME / len(pred_sep[n])\n", " RMSE = math.sqrt(RMSE / len(pred_sep[n]))\n", " stats.append([MAE,RME,RMSE,r_squared])\n", " MAE, RME, RMSE = 0, 0, 0\n", "\n", "print(stats)" ] }, { "cell_type": "code", "execution_count": 216, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "\n", "# Draw the graph for each room type\n", "# Red points are predictions and blue line is targeted price\n", "fig2, ax2 = plt.subplots(2, 3, figsize=(15, 10))\n", "ax2[0, 0].plot(org_sep[0], org_sep[0], label='Predicted', linewidth=2)\n", "ax2[0, 0].scatter(org_sep[0], pred_sep[0], color='r')\n", "ax2[0, 0].set_title('Studio')\n", "ax2[0, 1].plot(org_sep[1], org_sep[1], label='Predicted', linewidth=2)\n", "ax2[0, 1].scatter(org_sep[1], pred_sep[1], color='r')\n", "ax2[0, 1].set_title('Two rooms')\n", "\n", "ax2[0, 2].plot(org_sep[2], org_sep[2], label='Predicted', linewidth=2)\n", "ax2[0, 2].scatter(org_sep[2], pred_sep[2], color='r')\n", "ax2[0, 2].set_title('Three rooms')\n", "\n", "ax2[1, 0].plot(org_sep[3], org_sep[3], label='Predicted', linewidth=2)\n", "ax2[1, 0].scatter(org_sep[3], pred_sep[3], color='r')\n", "ax2[1, 0].set_title('Four rooms')\n", "\n", "ax2[1, 1].plot(org_sep[4], org_sep[4], label='Predicted', linewidth=2)\n", "ax2[1, 1].scatter(org_sep[4], pred_sep[4], color='r')\n", "ax2[1, 1].set_title('Five rooms')\n", "\n", "ax2[1, 2].plot(org_sep[5], org_sep[5], label='Predicted', linewidth=2)\n", "ax2[1, 2].scatter(org_sep[5], pred_sep[5], color='r')\n", "ax2[1, 2].set_title('Six or more')\n", "\n", "txt = '\\n'.join((\"MAE: {0:.0f} €\".format(stats[0][0]), \"RME: {0:.2f}%\".format(stats[0][1]), \"RMSE: {0:.0f} €\".format(stats[0][2]), \"R²: {0:.2f}\".format(stats[0][3])))\n", "ax2[1, 2].text(-2.6, 1.7, txt, fontsize=14,\n", " verticalalignment='bottom', transform=ax.transAxes)\n", "\n", "txt = '\\n'.join((\"MAE: {0:.0f} €\".format(stats[1][0]), \"RME: {0:.2f}%\".format(stats[1][1]), \"RMSE: {0:.0f} €\".format(stats[1][2]), \"R²: {0:.2f}\".format(stats[1][3])))\n", "ax2[1, 2].text(-1.4, 1.7, txt, fontsize=14,\n", " verticalalignment='bottom', transform=ax.transAxes)\n", "\n", "txt = '\\n'.join((\"MAE: {0:.0f} €\".format(stats[2][0]), \"RME: {0:.2f}%\".format(stats[2][1]), \"RMSE: {0:.0f} €\".format(stats[2][2]), \"R²: {0:.2f}\".format(stats[2][3])))\n", "ax2[1, 2].text(-0.2, 1.7, txt, fontsize=14,\n", " verticalalignment='bottom', transform=ax.transAxes)\n", "\n", "txt = '\\n'.join((\"MAE: {0:.0f} €\".format(stats[3][0]), \"RME: {0:.2f}%\".format(stats[3][1]), \"RMSE: {0:.0f} €\".format(stats[3][2]), \"R²: {0:.2f}\".format(stats[3][3])))\n", "ax2[1, 2].text(-2.6, 0.5, txt, fontsize=14,\n", " verticalalignment='bottom', transform=ax.transAxes)\n", "\n", "txt = '\\n'.join((\"MAE: {0:.0f} €\".format(stats[4][0]), \"RME: {0:.2f}%\".format(stats[4][1]), \"RMSE: {0:.0f} €\".format(stats[4][2]), \"R²: {0:.2f}\".format(stats[4][3])))\n", "ax2[1, 2].text(-1.4, 0.5, txt, fontsize=14,\n", " verticalalignment='bottom', transform=ax.transAxes)\n", "\n", "txt = '\\n'.join((\"MAE: {0:.0f} €\".format(stats[5][0]), \"RME: {0:.2f}%\".format(stats[5][1]), \"RMSE: {0:.0f} €\".format(stats[5][2]), \"R²: {0:.2f}\".format(stats[5][3])))\n", "ax2[1, 2].text(-0.2, 0.5, txt, fontsize=14,\n", " verticalalignment='bottom', transform=ax.transAxes)\n", "\n", "\n", "plt.xticks(fontsize=10)\n", "plt.yticks(fontsize=10)\n", "\n", "for ax in ax2.flat:\n", " ax.set(xlabel='Original price', ylabel='Predicted price')\n", "\n", "# Hide x labels and tick labels for top plots and y ticks for right plots.\n", "for ax in ax2.flat:\n", " ax.label_outer()" ] }, { "cell_type": "code", "execution_count": 186, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "c:\\users\\jussi kalliola\\appdata\\local\\programs\\python\\python36\\lib\\site-packages\\tensorflowjs\\converters\\keras_h5_conversion.py:122: H5pyDeprecationWarning: The default file mode will change to 'r' (read-only) in h5py 3.0. To suppress this warning, pass the mode you need to h5py.File(), or set the global default h5.get_config().default_file_mode, or set the environment variable H5PY_DEFAULT_READONLY=1. Available modes are: 'r', 'r+', 'w', 'w-'/'x', 'a'. See the docs for details.\n", " return h5py.File(h5file)\n" ] } ], "source": [ "#tfjs.converters.save_keras_model(best_model, './models/wandb/')\n", "#tfjs.converters.save_keras_model(model, './models/etuovi_model13/')\n", "#scalerDf.to_csv(r'.\\models\\wandb\\scaler.csv')" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.6.6rc1" } }, "nbformat": 4, "nbformat_minor": 2 }