Scribd
Upload
29 views9 pages

Stock Price Prediction Presentation

Uploaded by

Rishi Chourasia
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
29 views9 pages

Stock Price Prediction Presentation

Uploaded by

Rishi Chourasia
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 9

Stock Price

Prediction
using Time
Series and NLP

Training Seminar Report


Rishi Chourasia
21112089
Objective and
Problem Statement

Problem Statement:
• Stock price prediction is challenging due to
market volatility and the impact of multiple
external factors.
• Traditional models struggle with the complex,
non-linear patterns in financial data.

Objective:
• Use advanced models (ARIMA and LSTM) to
improve stock price prediction accuracy.
• Compare these models on the SPY ETF dataset
to identify the best forecasting approach.
Data Collection

Dataset: SPY ETF (Exchange Traded Fund)


Data Source: The data was collected from a reliable
source like Yahoo Finance and Alpha Vantage.
Data Points:
Open, close, high, low prices, and volume of
trades.
These features are important for analyzing stock
price movements.
Time Period: The dataset covers multiple years to ensure
robustness in identifying trends, including periods of
high volatility.
Data Features:
• Historical Prices: The key features are the open,
close, high, and low prices of the SPY ETF on a daily
basis.
• Trading Volume: Volume data is used to identify
market activity and investor sentiment
Data Preprocessing
Missing Values:Replace
missing values with
Why?: Preserves data 2> Outliers Detection:
1> Data Cleaning: median, and
size and avoids bias. Use IQR methods.
forward/backward fill
(for time series).

Moving Averages: Scaling: Min-Max scaling


3> Feature Engineering: SMA/EMA used to 4> Normalization: applied for model
smooth volatility. compatibility.

Data Split: 80% for


5> Train-Test Split:
training, 20% for testing
Model Used

1. Naive Forecast Model: Uses today’s price as tomorrow’s prediction,


serving as a baseline.
2. Moving Average Models: Simple Moving Average (SMA) and Exponential
Moving Average (EMA) capture trends over fixed windows.
3. ARIMA Model: A statistical model adept at handling linear relationships.
4. Recurrent Neural Networks (RNN): Maintains memory of sequential data;
LSTM (a type of RNN) captures long-term dependencies effectively.
Mean Absolute Error (MAE)
No Bias from Outliers: Unlike MSE, it Simplicity: Straightforward for model
doesn't heavily penalize large errors. comparison.

Model
Achieved MAE: ~1.2, indicating
Evaluation reasonable accuracy in predictions.

Accuracy: Lower MAE = better


performance.
Internship
Certificate
Thankyou

You might also like