A Comparative Study of Forecasting methods

Time series forecasting is the process of predicting the future value of a sequence based on historical data. Research in the area of machine learning has recently been focused on this task to address the limitations of traditional forecasting methods, which are time-consuming and complex. A powerful forecasting technique that is able to infer stochastic dependencies between past and future values is highly needed with the increase of historical data and the need for accurate production forecasting.

Introduction

In this project we studied various methods for predicting time series datas like stock market and comparing th results of these models

• Moving Average
• Prophet
• Long Short Term Memory (LSTM)

Problem details

Main Data:

Results

• Moving Avg:

  

• LSTM:

  

• Prophet:

  

Performance comparison

Comparing RMSE results of these models:

Method	RMSE
MV	8129.87
LSTM	809.013
Prophet	9292.317

Refrences:

[1] Shah, Dev, Wesley Campbell, and Farhana H. Zulkernine. "A comparative study of LSTM and DNN for stock market forecasting." 2018 IEEE International Conference on Big Data (Big Data). IEEE, 2018.

[2] Ariyo, Adebiyi A., Adewumi O. Adewumi, and Charles K. Ayo. "Stock price prediction using the ARIMA model." 2014 UKSim-AMSS 16th International Conference on Computer Modelling and Simulation. IEEE, 2014.

[3] Taylor, Sean J., and Benjamin Letham. "Forecasting at scale." The American Statistician 72.1 (2018): 37-45.

[4] Müller, K-R., et al. "Predicting time series with support vector machines." International Conference on Artificial Neural Networks. Springer, Berlin, Heidelberg, 1997.

[5] Gers, Felix A., Jürgen Schmidhuber, and Fred Cummins. "Learning to forget: Continual prediction with LSTM." Neural computation 12.10 (2000): 2451-2471.

Note

This implementation is based on Tensorflow 2.0 package and made possible by Google Colabratory GPU.