CsvStats.py

import os
import numpy as np
import pandas as pd
import pickle
from DataLoader import _filterDataframe
import  sys
"""
creates a csv file detailing the age/gender breakdown of the csv dataset

Params
    csvdata:    a pandas dataframe from the csv file we are using as out dataset
    ageRange:   the start and end values for the ages we are using
    outPath:    the path to save the stats csv at

Returns
    0: a pandas dataframe representing the results
"""
def statsCsv(csvdata, ageRange=[10, 100], outPath="stats.csv"):
    numRows = len(csvdata.index)
    numAges = ageRange[1] - ageRange[0] + 1
    resultsArr = np.zeros([numAges, 2], dtype=int)
    for i in range(numRows):
        sex = int(csvdata["isMale"][i])
        age = int(csvdata["age"][i])
        adjustedAge = age - ageRange[0]
        resultsArr[adjustedAge, sex] = resultsArr[adjustedAge, sex] + 1
    df = pd.DataFrame(resultsArr, columns=["female", "male"], index=np.arange(ageRange[0], ageRange[1]+1))
    df.to_csv(outPath)
    return df

"""
Finds the min and max resolution images in the dataset

Params
    csvData:    a pandas dataframe from the csv file we are using as out dataset

Returns
    0:  a string containing information about the min and max res images
"""
def findImageSizeRange(csvData):
    numRows = len(csvData.index)
    minRes = float("inf")
    maxRes = 0
    minResStr = ""
    maxResStr = ""
    sumRes = 0
    for i in range(numRows):
        width = csvData["image_width"][i]
        height = csvdata["image_height"][i]
        resStr = "[" + str(width) + ", " + str(height) + "]"
        res = height * width
        sumRes = sumRes + res

        if res > maxRes:
            maxRes = res
            maxResStr = resStr
        if res < minRes:
            minRes = res
            minResStr = resStr
        if i % 100000 == 0:
            print (str(i) + "/" + str(numRows))
    return "min:" + minResStr + " max:" + maxResStr

if __name__ == "__main__":
    if len(sys.argv) == 3:
        csvPath = sys.argv[1]
        indicesPath = sys.argv[2]

        if os.path.exists(csvPath) and os.path.exists(indicesPath):
            print("restoring csv data...")
            csvdata = pd.read_csv(csvPath)

            print("restoring indices data...")
            file = open(indicesPath, "rb")
            indices = pickle.load(file)
            file.close()

            print ("finding image size range...")
            imgRange = findImageSizeRange(csvdata)
            print (imgRange)
            print("generating stats csv...")
            statsCsv(csvdata)

        else:
            print("one or both files not found")
    else:
        print("requires 2 parameters (csv_path, indices_path)")