stanford997 / cs777-term-paper

1. Set up environment

1.1 python

pwd

python3 --version

python 3 -m venv py_env

source py_env/bin/activate

1.2 airflow

1.2.1

pip install 'apache-airflow==2.8.3' \
 --constraint "https://raw.githubusercontent.com/apache/airflow/constraints-2.8.3/constraints-3.8.txt"

curl -LfO 'https://airflow.apache.org/docs/apache-airflow/2.8.3/docker-compose.yaml'

mkdir -p ./dags ./logs ./plugins ./config

docker compose up airflow-init

1.2.2

in docker-cmopose.yaml replace image:

and add ports : 5432:5432 to map container's internal port 5432 to port 5432 on the host machine.

# image: ${AIRFLOW_IMAGE_NAME:-apache/airflow:2.8.3}
image: ${AIRFLOW_IMAGE_NAME:-extending_airflow:latest}

  postgres:
    image: postgres:13
    environment:
      POSTGRES_USER: airflow
      POSTGRES_PASSWORD: airflow
      POSTGRES_DB: airflow
    volumes:
      - postgres-db-volume:/var/lib/postgresql/data
    ports:
      - 5432:5432
    healthcheck:
      test: [ "CMD", "pg_isready", "-U", "airflow" ]
      interval: 10s
      retries: 5
      start_period: 5s
    restart: always

dockerfile

FROM apache/airflow:2.8.3
COPY requirements.txt /requirements.txt
RUN pip install --user --upgrade pip
RUN pip install --no-cache-dir --user -r /requirements.txt

requirements.txt

pandas==2.0.3
yfinance==0.2.31
psycopg2-binary
scikit-learn==1.3.2
numpy==1.23.5
keras==2.13.1
tensorflow==2.13.1

docker build . --tag extending_airflow:latest

This command builds a Docker image named extending_airflow and tags it as latest. Specifically, it searches for a file named Dockerfile in the current directory and follows the instructions provided in that file to build the image. During the build process, each command in the Dockerfile, such as FROM, COPY, RUN, etc., is executed to create the image with the required environment and application. Once the build is complete, the resulting image is labeled with the specified tag, which can be latest to indicate it as the most recent version.

1.2.3

docker compose up -d

docker ps

1.3 minio

mkdir -p ~/minio/data

docker run \
   -p 9000:9000 \
   -p 9001:9001 \
   --name minio \
   -v ~/minio/data:/data \
   -e "MINIO_ROOT_USER=ROOTNAME" \
   -e "MINIO_ROOT_PASSWORD=CHANGEME123" \
   quay.io/minio/minio server /data --console-address ":9001"

1.4 create connection using Airflow UI

1.4.1 postgres connection

1.4.2 minio connection

2. Run

Go to http://localhost:8080/ Airflow UI and DAGs.

Then run it.

3. DAG

task1 >> task2 >> model_sensor >> [task3, task4, task5] >> task6

task1 - query_stock_info:

Connect to local database -> create table if not exist -> get and save stock data in local postgres database(table name: stock_info)

task2 - processed_stock_data:

Save the projection of column ['Stock_Name', 'Date', 'Open', 'Close'] to table 'processed_stock_data'

model_sensor - check_models_in_minio:

check if the models in minio, if not wait until it is saved in minio. (Timeout, poke_interval)

task3, 4, 5 - 3 models:

Load models from minio

Each one will give the results 1 or 0, 1 means good stock, 0 means bad

task6 - make_investment_decision:

Make investment decision based on Task 3,4,5

Only if the prediction results of more than or equal to two models are good stocks, the final result is a good stock.

4. Dataset and Results

4.1 dataset

Our program will download the specified stocks dataset from Yahoo. Here is a screen shot of dataset:

stock_name: Stock name in Yahoo, apple -> AAPL, Nvdia -> NVDA

date: date of the stock

open: open price is the price at which a particular stock starts trading when the stock market opens for the day

high: the highest price of the stock in a day

low: the lowest price of the stock in a day

close: close price is the final price at which a particular stock is traded on a given trading day

volume: the total number of shares or contracts traded

adj_close: adjusted close price is the final price adjusted before the next trading day

short_ma: short-term moving average of a financial metric

long_ma: long-term moving average of a financial metric

4.2 Results

SELECT psd.date,
       psd.stock_name,
       psd.open,
       psd.close,
       sp.prediction,
       (psd.close - psd.open) AS earning
FROM processed_stock_data psd
JOIN stock_prediction sp ON psd.date = sp.date AND psd.stock_name = sp.stock_name
WHERE psd.stock_name = 'NVDA';

You can use this SQL to see your stock results.

Below are some examples of our results with stock NVDA,

4.2.1 NVDA

3 correct, 6 wrong predictions

4.2.2 GOOGL

6 correct, 3 wrong predictions

4.2.3 AMZN

5 correct, 4 wrong predictions

5. Run on Google Cloud

5.1 create composer(airflow on GC)

https://console.cloud.google.com/composer/

Create composer2.

You can add python packages using PYPI. Then composer will download for you automatically.

5.2 create postgres database

You need to add your airflow ip to authorized networks here.

And create connection in Airflow UI using the postgres database public IP address.

5.3 Google Cloud Storage

Here allow public access to the bucket and save the models in the bucket.

gcs_hook = GoogleCloudStorageHook(google_cloud_conn_id)
# First download .h5 model file from minio and then load model
model_stream = gcs_hook.download(bucket_name=bucket_name, object_name=model_key, filename=f'/tmp/{model_key}')

Because we use GoogleCloudStorageHook to download the model.

5.4 Composer Dashboard

Logs:

Dags:

Monitor: