Skip to content

Latest commit

 

History

48 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SC1015 Data Science Mini-Project - Stocks

About

In this repository, we will be attempting to find the top factors that affect the performance of a stock, so as to develop a safe stock investing strategy to ensure positive rates of return for investors while minimizing losses. We will be mainly focusing on the US Stock Market data taken from kaggle

Models used

  • Decision Tree
  • Random Forest
  • GridSeach Cross-Validation
  • Time series

Folder descriptions

Contains:- 
- 2 CSV files (before and after cleaning)
- Python Notebook for data cleaning

In this folder, we are doing basic data preparation before moving on to analysis. 
This includes filling Nan values in the dataset and removing outliers.
Contains:-
- 2 CSV files (before and after EDA)
- Python Notebook for EDA

In this folder, we will be exploring the dataset. We looked into the importance of columns "Price Var" and "Class". 
Realising that there are too many factors to work with, we cut down that number down into a workble size.
Utilizing correlation heatmaps, boxplots and violin plots, we explored the factors and found the most relavant factors
to be used in our machine learning model.
Contains:-
- Dataset after EDA in CSV format
- Python Notebook for Building the Model
- Model in sav format

In this folder, we are utilizing decision tree, random forest and GridSearch to create a machine learning model which
can help us predict the "Class" of a stock. This would let us know whether a stock is profitable or not.
In the notebook, we explore the various models to look at how do we improve our classification accuracy.
Contains:-
- Model in sav format
- 4 CSV files containing datasets from different years to test our model on
- Python Notebook for looking into what strategy should we consider to maximize profit while minimizing losses.

In this folder, we will be using our model to predict the "Class" variables of the various stocks in different years.
We then proceeded to propose 3 different strategies for investing based on our model, ran a simulation and plotted 
the results in a time series graph. After some analysis of the time series graph, we obtained the best strategy for 
an investor to maximize profit while minimizing losses.

Additional materials included

  • Slides for the presentation

Conclusion

  • We have developed a model that utilizes various factors that are readily available on trading platforms to predict and classify a stock into its proper class
  • We also utilized a time series graph to help us simulate our proposed strategy
  • With our model and proposed strategy, it will allow investors to make more informed decision on which stock to invest in to allow for profits while minimizing losses.

Contributors

SC16, Group 7

  • Chantharojwong Kasidis - Data Preparation + EDA + Strategy
  • Harvey Zhang Tianren - Presenter + Editor
  • Yeoh Wei Yang - Model building + Slides

References

MISC.

This repository is submitted as a project work for Nanyang Technological University's SC1015- Data Science and Aritificial Intelligence course.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages