Sequence time series model

Hi,

my name is fatihah and I am currently develop RNN model to forecast energy consumption for residential buildings from historical 7 years data.
However, I have a problem to plot the data, it produce cluster as shown below:


Anyone know what it could be?
the data in my csv file is just as shown below:
ENERGY_CONSUMPTION_CURRENT,INSPECTION_DATE

First data point looks like this:

242.0,2022-11-17

Second data point looks like this:

101.0,2022-10-01
my reference is C4W4 assignment 4 from the sequence time series courses.

You should have one point per data, I dont know how this plot was made, the first here:

242.0,2022-11-17 should be the kW and then the data!

It could be due to the fact that the 2 vectors, time and value, aren’t aligned the way they should be in your plot.
Try plotting the data as points, as supposed to line, and see how it looks. Then plot the first 10 points, and make sure they are associated the right way, and ordered the right way. The points should be ordered by time, i.e. the second point should come after the first point in time. So on so forth. And go from there.

does this seems okay?


this is my final output with
mse: 84.93, mae: 6.62 for forecast
I should tuning for training size,window size and learning rate to reduce mse & mae right?

2 Likes