Prediction of Stock Price Movements Using Statistical and Hybrid Regression Techniques to Reduce Diversify Risk

Singh B., Henge S. K.

4th International Conference on Intelligent and Fuzzy Systems (INFUS), Bornova, Türkiye, 19 - 21 Temmuz 2022, cilt.505, ss.456-462 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası: 505
  • Doi Numarası: 10.1007/978-3-031-09176-6_52
  • Basıldığı Şehir: Bornova
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.456-462
  • Anahtar Kelimeler: Gradient Boost (GB), Random Forest (RF), Bagging Regression (BR), MLP-Regression (MLP-R), Ad-Boost (AB), Naive Bayes (NB)
  • İstanbul Teknik Üniversitesi Adresli: Hayır

Özet

The situation of Over-Fitting occurs when model tries to study the features of data in details and found some noise in data to some extent in training phase that hamper the performance of the model in new upcoming data. This paper emphasised on over fitting problems for long interval continuous series of data that makes investors to make decision complexity. Data has been calibrated through yahoo finance for Forecasting of banking stock named HDFC bank and State Bank of India. The study also successfully identifies the best parameters set for better Return on Investment using super trend parameter optimization. Second Part of the paper is to access the simulation of risk management for long term investment. Third part of research mainly focused on implementing regression and classification for predicting future price of banking stock. Best Accuracy evaluation for Linear regression come out to be 0.971 and MLP Regressor is 0.973. WCM, ROI and R-2 Score are used for the best return investments using super trend parameter based optimization.