How to tune parameters for Time Series Analysis, when forecasting is only dominated by one feature and error is not getting reduced?

I am trying to predict time series based on 150 features. When I plot correlation of these features, I am getting 20 features with more or less importance but every model I use, it is completely dominated by only one feature which is competently in sync with predicted output but not actual output . Please refer to the image below.

The green line is prediction which is completely in sync with one of the feature.And for every valley in actual output, I am getting 2 valleys in predicted output. No model is able to generalize this. Is it the case of bad data for model?

Topic hyperparameter deep-learning time-series feature-extraction machine-learning

Category Data Science

1
answered at

The current method of XGboost and tree ensemble appears to not be closely modeling the actual times-series pattern. It might be better to use times-series-centric algorithms like AutoRegressive Integrated Moving Average (ARIMA) or Long Short Term Memory (LSTM).

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.