# Time series with python

I've a dataset in following format

day,8am,9am,10am,......,8pm
2012-1-1,0.3,0.2,0.21,.....,0.4
2012-1-2,0.32,0.19,0.22,.....,0.45
................


I'm not sure whether I processed the whole dataset and get a single time series like follows by combining each day

0.3,0.2,0.21,.....,0.4,0.32,0.19,0.22,.....,0.45,........


or is there any better way to analyze such a dataset?

I'm not interested in day. I just trying to focus on the values

• This questions is quite unclear in its current form. Can you provide more information? What do 'val1','val2', etc correspond to? Are they the same measurement, just at $N$ different times of the day? Or are they for example readings of different sensors on the same entity? – deemel Mar 18 at 12:30
• @Rickyfox they are hourly measurements of the electricity usage. e.g. 8 am, 9 am, 10 am, etc.. – Thusitha Thilina Dayaratne Mar 18 at 12:32
• Then you should replace 'day' with 'hour', resulting in a series of ('hour','measurement') tuples. – deemel Mar 18 at 12:40
• that would be convenient for me as I would build a daily total model and then use daily totals to predict hourly usage and then reconcile the daily predictions leading to a forecast for each hour for the next k days. Please post your data as a matrix of 24 columns and k days (rows). – IrishStat Mar 18 at 13:35
• your format is correct for your purposes. The format I suggested would be more amenable to my approach. Present both here. – IrishStat Mar 19 at 8:14