Challenges to Time Series Analysis
The client gave us the data which was similar to a sensor data or Stock Market Data.
-
The data consisted of ~250 variables ‘col1’,’col2’...etc. And the data size was 15gb.
-
Besides this, there are two important categorical variables named ‘symbol,' ‘categ’ and another variable ‘time.'
-
There are 22 unique symbols, and seven unique categ’s in the whole dataset which has recorded for every minute.
-
The target is to forecast 10 future values of column named ‘val’ for each symbol-categ pair.
Solution Offered for Time Series Analysis and Forecasting
We offered two different approaches for the problem statement of the client.
-
Approach 1 - Convert Time Series Problem to Supervised Learning Problem.
In this approach, we consider converting Time-Series data to Supervised Learning data. In Supervised Learning, we require the values for all the independent variables.
-
Approach 2 - Using VAR(Vector Autoregression) Model.
This is an extension to the one dimensional Autoregressive Method. The advantage of this method is, we need not convert the Time Series Data to Supervised Learning Data.
We can get the predictions for future values from the model itself. The model considers the interdependencies in the data.