1. distribuir na internet

Stock Data Analysis (assets_mon.py)

This program is a class for stock data that uses the yahoo_fin library to pull stock data for a given symbol. The class has several methods, including:

• get_last_day: returns data for the last trading day.
• get_live_data: returns the live price of the stock.
• get_history: returns historical data for the stock.
• get_prev: returns a prediction of the next day's closing price for the stock using pre-trained models.

The script loads pre-trained models specific to different stock symbols. The get_prev method checks the symbol of the stock and loads the corresponding pre-trained model to make a prediction for the next day's closing price. It also drops specific columns from the last day's data to prepare it for the prediction.

The script also uses the keras library for creating and loading models. It also uses the pandas library for data manipulation.

Installation

stock = Stock('PETR4.SA')To install the dependencies from the requirements.txt file, navigate to the root directory of the project and run the following command:

Copy code
pip install -r requirements.txt

This will install all the packages listed in the requirements.txt file in your environment.

and It is recommended to use a virtual environment for this project, you can use virtualenv or conda to create a virtual environment and then activate it, then install the dependencies.

Usage

To use the class, you should import the class Stock and create an instance of the class passing the stock symbol as an argument.

You could try in a virtual env, importing everthing to python3 prompt.

import assets_mon as ats
# stock object 
stock = ats.Stock('PETR4.SA')
# get_last_day() returns open, high, close, adj close and volume
print(stock.get_last_day())
# get_live_data() returns the price with some delay 
print(stock.get_live_data())
# get_history() returns all times historical data
print(stock.get_history())
# get_prev() return a prevision for the next day closing price
print(stock.get_prev())

For Brazilians stocks market you should use the stock symbol plus ' .sa ' -> ' petr4.sa ', this is needed to differentiate from others markets.

For American market you just need the stock symbol. Example -. ' aappl '

Please note that the pre-trained models are specific to the stocks symbols in the "stocks.txt" file located in the models directory. You can also use the class to get data from other stocks, but the prediction will not be available.

Please also note that the data returned by the methods are in the form of a pandas DataFrame, so you can use the pandas library to manipulate the data as you see fit.

Stocks Monitor (monitor.py)

​ This is a Python example program that allows the user to input some stocks and retrieve their future closing price for the next trading day. The results are displayed in a rich table and also stored in a JSON file.

Requirements

• Python 3.x

• rich library

• pandas library

• assets_mon library

Usage

​ A list of stocks symbols is stored in a text file (./models/stocks.txt). The program reads the stocks symbols from the file and inputs each one to retrieve its future closing price.

​ The resulting rich table displays the stock symbol and the future closing price for each stock.

​ The resulting JSON file stores the stock symbol and the future closing price for each stock in JSON format. The file name is saved in ./json/stocks_json_dd-mm-yyyy.json.

How to Run

1. Make sure you have installed the required libraries.

Copy code
pip install -r install.txt

3. Run the monitor.py file.

Copy code
python monitor.py

1. The results will be displayed in a rich table and also stored in a JSON file.

LSTM Stocks Model Generator (prevision_futures.py)

This code is for predicting the future stock prices using a LSTM neural network. The data used for training the model is the stock information for the past 5 years obtained using the yahoo_fin library.

Requirements

• yahoo_fin
• pandas
• numpy
• scikit-learn
• keras

Copy code
pip install -r install.txt

Inputs

​ The input to the prevision_futures function is a single string representing the stock symbol for which the future prices need to be predicted.

Outputs

The output of the function is a saved model in the ./models directory with the file name <stock symbol>_.h5.

Features

The features used for the stock price prediction are:

• Open price
• High price of the day
• Low price of the day
• Adj Close in the day
• Volume in the day

Model Architecture

The model architecture consists of:

• 2 LSTM layers with 128 and 64 units respectively
• 2 dense layers with 25 and 1 units respectively
• The model uses the Adam optimizer and the mean squared error as the loss function.

Training

The model is trained on 80% of the data and validated on 20% of the data. The model is trained for 30 epochs.

Know Issues

1. Sometimes JSON files not work properly, so please go to https://jsonformatter.curiousconcept.com/# and upload the json file, it should resolve this issue.

2. In case if you want to test prevision_futures.py use a GPU (Graphic Processing Unit) with 500 CUDAs or more, it also work with CPU but sometimes the result is not very good and there is a loss in processing data, so it might gave some bias or wrong results.