Analysis of BMRI Stock Prices Using Python and LTSM in Google Colab

2026-07-19 · Source: Machine Learning on Medium · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Capital Markets & Investment Management · Depth: Intermediate, extended

Summary

A project analyzed and predicted PT Bank Mandiri (Persero) Tbk (BMRI) stock prices using Python and a Long Short-Term Memory (LSTM) model in Google Colab. Historical data from Yahoo Finance, spanning January 1, 2022, to December 8, 2025, was retrieved and preprocessed using MinMaxScaler. An LSTM model, comprising two 50-neuron layers and a Dense output layer, was trained for 10 epochs with a batch size of 32, achieving a Mean Squared Error (MSE) of 24,017.37. The model then predicted BMRI stock prices for the next 30 days, showing a gradual upward trend. These predictions were subsequently analyzed by a Llama 3 Large Language Model via Ollama, which interpreted the trends and offered recommendations, demonstrating an integrated approach for financial forecasting.

Key takeaway

For data scientists building financial forecasting systems, this integrated approach demonstrates how to combine robust time series prediction with interpretive AI. You should consider using LSTM for its pattern recognition capabilities on historical stock data and then integrate an LLM like Llama 3 to generate human-readable analyses and recommendations, enhancing the utility of raw predictions for stakeholders. This can streamline your workflow from data to actionable insights.

Key insights

Integrating LSTM for stock prediction with LLM for analysis provides comprehensive financial forecasting.

Principles

• Historical closing prices are key for stock prediction.
• Normalization improves model learning stability.
• Sliding window method forms time series data.

Method

Retrieve historical stock data, normalize closing prices using MinMaxScaler, train an LSTM model with a sliding window (n_steps=60), then use iterative forecasting for future predictions. Integrate with an LLM for interpretive analysis.

In practice

• Use "yfinance" for direct stock data access.
• "mplfinance" visualizes candlestick charts effectively.
• "MinMaxScaler" prepares data for neural networks.

Topics

• LSTM
• Stock Price Prediction
• Large Language Models
• Financial Forecasting
• Python Libraries
• Ollama

Best for: Machine Learning Engineer, Data Scientist, AI Student

Related on AIssential

• Representing Time Series as Structured Programs for LLM Reasoning — Representing time series as structured symbolic programs enables LLMs to reason effectively without fine-tuning.
• Beyond Polarity: Multi-Dimensional LLM Sentiment Signals for WTI Crude Oil Futures Return Prediction — Multi-dimensional LLM sentiment, beyond polarity, improves WTI crude oil futures return prediction.
• Bridging the Last Mile of Time Series Forecasting with LLM Agents — LLM agents can bridge the gap between statistical time series forecasts and business-ready predictions by integrating weakly structured contextual information.
• Building AI for Stock Price Prediction with LSTM: A Hands-on Record of Deep Learning from Scratch — Rigorous deep learning for stock prediction requires robust evaluation beyond simple metrics, often confirming market efficiency.
• Speaking Numbers to LLMs: Multi-Wavelet Number Embeddings for Time Series Forecasting — TempoWave improves LLM time series forecasting by aligning continuous numerical data with discrete language models using multi-wavelet embeddings.
• Introduction to quantitative finance Part 26: Whether to use forecasting methods, or to tell an LLM… — Forecasting in quantitative finance faces challenges with LLMs, necessitating memory-enhanced reward shaping for consistency.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning on Medium.