keras_npy.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2021-02-14T07:12:18.561887Z",
     "iopub.status.busy": "2021-02-14T07:12:18.560934Z",
     "iopub.status.idle": "2021-02-14T07:12:18.564081Z",
     "shell.execute_reply": "2021-02-14T07:12:18.563465Z"
    },
    "papermill": {
     "duration": 0.017173,
     "end_time": "2021-02-14T07:12:18.564201",
     "exception": false,
     "start_time": "2021-02-14T07:12:18.547028",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-14T07:12:18.588665Z",
     "iopub.status.busy": "2021-02-14T07:12:18.587838Z",
     "iopub.status.idle": "2021-02-14T07:12:24.414517Z",
     "shell.execute_reply": "2021-02-14T07:12:24.412886Z"
    },
    "papermill": {
     "duration": 5.842287,
     "end_time": "2021-02-14T07:12:24.414649",
     "exception": false,
     "start_time": "2021-02-14T07:12:18.572362",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import numpy as np # linear algebra\n",
    "import pandas as pd\n",
    "from tensorflow.keras.layers import Input, Dense, BatchNormalization, Dropout, Concatenate, Lambda, GaussianNoise, Activation\n",
    "from tensorflow.keras.models import Model, Sequential\n",
    "from tensorflow.keras.losses import BinaryCrossentropy\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.callbacks import EarlyStopping\n",
    "from tensorflow.keras.layers.experimental.preprocessing import Normalization\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import GroupKFold\n",
    "\n",
    "from tqdm import tqdm\n",
    "from random import choices\n",
    "import random\n",
    "\n",
    "import kerastuner as kt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-14T07:12:24.446125Z",
     "iopub.status.busy": "2021-02-14T07:12:24.445102Z",
     "iopub.status.idle": "2021-02-14T07:14:55.162355Z",
     "shell.execute_reply": "2021-02-14T07:14:55.161108Z"
    },
    "papermill": {
     "duration": 150.739524,
     "end_time": "2021-02-14T07:14:55.162514",
     "exception": false,
     "start_time": "2021-02-14T07:12:24.422990",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "train = pd.read_csv('../input/jane-street-market-prediction/train.csv')\n",
    "train = train.query('date > 85').reset_index(drop = True) \n",
    "train = train.astype({c: np.float32 for c in train.select_dtypes(include='float64').columns}) \n",
    "train = train.query('weight > 0').reset_index(drop = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-14T07:14:55.185149Z",
     "iopub.status.busy": "2021-02-14T07:14:55.183631Z",
     "iopub.status.idle": "2021-02-14T07:14:55.185880Z",
     "shell.execute_reply": "2021-02-14T07:14:55.186378Z"
    },
    "papermill": {
     "duration": 0.015865,
     "end_time": "2021-02-14T07:14:55.186518",
     "exception": false,
     "start_time": "2021-02-14T07:14:55.170653",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "features = [c for c in train.columns if 'feature' in c]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-14T07:14:55.453462Z",
     "iopub.status.busy": "2021-02-14T07:14:55.452507Z",
     "iopub.status.idle": "2021-02-14T07:14:57.866829Z",
     "shell.execute_reply": "2021-02-14T07:14:57.868030Z"
    },
    "papermill": {
     "duration": 2.673456,
     "end_time": "2021-02-14T07:14:57.868239",
     "exception": false,
     "start_time": "2021-02-14T07:14:55.194783",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "CACHE_PATH = './'\n",
    "df = train[features]\n",
    "f_mean = df.median()  # df.mean()\n",
    "f_mean = f_mean.values\n",
    "np.save(f'{CACHE_PATH}/keras_f_mean_online.npy', f_mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-14T07:14:57.903543Z",
     "iopub.status.busy": "2021-02-14T07:14:57.902619Z",
     "iopub.status.idle": "2021-02-14T07:15:01.616893Z",
     "shell.execute_reply": "2021-02-14T07:15:01.615753Z"
    },
    "papermill": {
     "duration": 3.736818,
     "end_time": "2021-02-14T07:15:01.617036",
     "exception": false,
     "start_time": "2021-02-14T07:14:57.880218",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "train.fillna(train.median(),inplace=True)\n",
    "train['action'] =  (  (train['resp_1'] > 0 ) & (train['resp_2'] > 0 ) & (train['resp_3'] > 0 ) & (train['resp_4'] > 0 ) &  (train['resp'] > 0 )   ).astype('int')\n",
    "resp_cols = ['resp_1', 'resp_2', 'resp_3', 'resp', 'resp_4']\n",
    "X = train[features].values\n",
    "y = np.stack([(train[c] > 0).astype('int') for c in resp_cols]).T "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-14T07:15:01.642108Z",
     "iopub.status.busy": "2021-02-14T07:15:01.640707Z",
     "iopub.status.idle": "2021-02-14T07:15:01.645299Z",
     "shell.execute_reply": "2021-02-14T07:15:01.644572Z"
    },
    "papermill": {
     "duration": 0.01935,
     "end_time": "2021-02-14T07:15:01.645429",
     "exception": false,
     "start_time": "2021-02-14T07:15:01.626079",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1571415, 130)\n"
     ]
    }
   ],
   "source": [
    "print(X.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-14T07:15:01.668628Z",
     "iopub.status.busy": "2021-02-14T07:15:01.667863Z",
     "iopub.status.idle": "2021-02-14T07:15:01.673262Z",
     "shell.execute_reply": "2021-02-14T07:15:01.672690Z"
    },
    "papermill": {
     "duration": 0.018099,
     "end_time": "2021-02-14T07:15:01.673373",
     "exception": false,
     "start_time": "2021-02-14T07:15:01.655274",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1571415, 5)\n"
     ]
    }
   ],
   "source": [
    "print(y.shape)"
   ]
  }
 ],
 "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": "3.8.6"
  },
  "papermill": {
   "duration": 167.637983,
   "end_time": "2021-02-14T07:15:01.910351",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2021-02-14T07:12:14.272368",
   "version": "2.1.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}