Week 12

For all of the models we are reffering from this point onwards are trained using 80% of the data and the remaining 20% is used for testing

First we made a model which predicts the water level of kalu ganga one hour ahead. The model weights are shown below

Here the one hour lag of Kalu ganga seems to have the greatest impact on the prediction. one of the interesting observation is that the impact of wey ganga is very small even though it has the best correlation coefficient

The predicted curve and the actual curve is shown in the below figure

The prediction is following the actual plot to a good extent. The root mean square error is 13.64 cm

The boxplot and the error variation is shown below

It is oblivious that a 1 hour is not sufficient to communicate flood to the people and evacuate them. Therefore, we try to predict more hours ahead although error kept increased.

Page updated

Google Sites

Report abuse