initial commit, add readme, FFT

2018-11-05 14:58:15 -05:00
commit 3adbe95601
@@ -0,0 +1,4 @@
+.ipynb_checkpoints/
+*.dat
+*.npy
+
@@ -0,0 +1,166 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Valence value regression based on Deap Dataset\n",
+    "## 1. FFT with pyeeg\n",
+    "* [4-8]: theta band\n",
+    "* [8-12]: alpha band\n",
+    "* [12-16]: low beta band \n",
+    "* [16-25]: high beta band\n",
+    "* [25-45]: gamma band"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import numpy as np\n",
+    "import pyeeg as pe\n",
+    "import pickle as pickle\n",
+    "import pandas as pd"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 2. Channel Indice \n",
+    "File Name data\\SXX.dat, XX \\in [0,31]\n",
+    "* data: 40 x 40 x 8064: trial x channel x data\n",
+    "* label: 40 x 4: video/trial x label (valence, arousal, dominance, liking)\n",
+    "\n",
+    "Channel Indice: {\n",
+    "* 1 : AF3\n",
+    "* 2: F3\n",
+    "* 3: F7\n",
+    "* 4: FC5\n",
+    "* 7: T7\n",
+    "* 11: P7\n",
+    "* 13: O1\n",
+    "* 17: AF4\n",
+    "* 19: F4\n",
+    "* 20: F8\n",
+    "* 21: FC6\n",
+    "* 25: T8\n",
+    "* 29: P8\n",
+    "* 31: O2 }\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 62,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "channel = [1,2,3,4,6,11,13,17,19,20,21,25,29,31] #14 Channels chosen to fit Emotiv Epoch+\n",
+    "band = [4,8,12,16,25,45] #5 bands\n",
+    "window_size = 256 #averaging band power of 2 sec\n",
+    "step_size = 16 #each 0.125 sec update once\n",
+    "sample_rate = 128 #sampling rate of 128 Hz\n",
+    "\n",
+    "def FFT_Processing (sub, channel, band, window_size, step_size, sample_rate):\n",
+    "    '''\n",
+    "    arguments:  string subject\n",
+    "                list channel indice\n",
+    "                list band\n",
+    "                int window size for FFT\n",
+    "                int step size for FFT\n",
+    "                int sample rate for FFT\n",
+    "    return:     void\n",
+    "    '''\n",
+    "    \n",
+    "    meta = []\n",
+    "    with open('data\\s' + sub + '.dat', 'rb') as file:\n",
+    "\n",
+    "        subject = pickle.load(file, encoding='latin1') #resolve the python 2 data problem by encoding : latin1\n",
+    "\n",
+    "        for i in range (0,40):\n",
+    "            # loop over 0-39 trails\n",
+    "            data = subject[\"data\"][i]\n",
+    "            labels = subject[\"labels\"][i]\n",
+    "            start = 0;\n",
+    "\n",
+    "            while start + window_size < data.shape[1]:\n",
+    "                meta_array = []\n",
+    "                meta_data = [] #meta vector for analysis\n",
+    "                for j in channel:\n",
+    "                    X = data[j][start : start + window_size] #Slice raw data over 2 sec, at interval of 0.125 sec\n",
+    "                    Y = pe.bin_power(X, band, sample_rate) #FFT over 2 sec of channel j, in seq of theta, alpha, low beta, high beta, gamma\n",
+    "                    meta_data = meta_data + list(Y[0])\n",
+    "\n",
+    "                meta_array.append(np.array(meta_data))\n",
+    "                meta_array.append(labels)\n",
+    "\n",
+    "                meta.append(np.array(meta_array))    \n",
+    "                start = start + step_size\n",
+    "        meta = np.array(meta)\n",
+    "        np.save('out\\s' + sub, meta, allow_pickle=True, fix_imports=True)\n",
+    "\n",
+    "\n",
+    "subjectList = ['01','02','03','04','05','06','07','08','09','10','11','12','13','14','15','16','17','18','19','20','21','22','23','24','25','26','27','28','29','30','31','32']\n",
+    "\n",
+    "for subjects in subjectList:\n",
+    "    FFT_Processing (subjects, channel, band, window_size, step_size, sample_rate)"
+   ]
+  },
+  {
+   "attachments": {},
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## 3.Segment and training of preprocessed data\n",
+    "* training dataset\n",
+    "* testing dataset\n",
+    "* validation dataset\n",
+    "\n",
+    "Agrithom pool:\n",
+    "* Support Vector Machine (which kernal?)\n",
+    "* Ada-Boost\n",
+    "\n",
+    "Best practice could be refered to this paper: \n",
+    "\n",
+    "@article{alarcao2017emotions,\n",
+    "  title={Emotions recognition using EEG signals: A survey},\n",
+    "  author={Alarcao, Soraia M and Fonseca, Manuel J},\n",
+    "  journal={IEEE Transactions on Affective Computing},\n",
+    "  year={2017},\n",
+    "  publisher={IEEE}\n",
+    "}"
+   ]
+  },
+  {
+   "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": "3.6.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}
@@ -0,0 +1,16 @@
+# EEG Classifier based on DEAP database
+
+## This repo only include code to proceed DEAP data, no data from DEAP contained!
+
+For access of DEAP dataset, please sign EULA and send a request to DEAP team: http://www.eecs.qmul.ac.uk/mmv/datasets/deap/download.html
+
+## Dependency
+*python >= 3.5
+*numpy
+*pyEEG: https://github.com/forrestbao/pyeeg
+
+
+## ERD/ERS analysis
+Fast Fourier Transformation:
+*Windows size = 2 sec
+*Windows step = 0.125 sec