Friday, 23 December 2016

Electracker - More energy consumption analysis

Now that I have a few weeks of data, I wanted to get daily graphs to look for patterns. I amended the original code to create them:

Not as useful as I'd hoped, but the average consumption per day is definitely interesting.

The coding is pretty terrible, sorry. If I was starting from scratch I'd use datetime module functions to do the date work and include some error handling to stop the program crashing when data is missing for an hour. Because I was pushed for time and had a working program already in place  to modify, it was the quickest solution.

Code, such as it is,  is here:

#program to analyse electricity use over time by hour.

#open a file

#for each hour:
#read it line by line
#bucket into hour
#get total
#get average for each hour

#enhancements to follow: daily charts, date range charts, movie!

import csv
import os
import matplotlib
import matplotlib.pyplot as plt

startDay = 20
startMonth = 12

while True:

 for i in range (0,24):     #for each hour
    with open ('feeds.csv', 'rb') as csvfile:
     rawelecdata = csv.DictReader(csvfile)

     for row in rawelecdata:
      if int(row['created_at'][11:13])==i and int(row['created_at'][8:10])==startDay and int(row['created_at'][5:7])==startMonth:
       hourelectot += float(row['field1'])
       j +=1
    print(i,hourelectot/j,j) #useful to for debugging

    elecByHour.append(hourelectot/j) #add average for the current hour to list

    #calculate average for day:
    dailyTotal = dailyTotal + hourelectot
    dailyReadings = dailyReadings + j

 dailyAverage = dailyTotal/dailyReadings
#plot the graph:

 y = (elecByHour)
 N = len(y)
 x = range(N)
 width = 1/1.5
 plt.xlabel ('Time of Day / hr')
 plt.ylabel ('kW')
 plt.title('Consumption for '+ str(startDay) + '/' + str(startMonth) + ' Average = ' + str(dailyAverage)+'kW'), y, width, color="blue")
 plt.savefig('elecByHour' + str(startMonth) + str(startDay) + '.png')
 #os.system('xviewer elecByHour.png &')
 startDay += 1

No comments:

Post a Comment