Team Warwick

• 

  Actual vs Augmented ARIMA Predicted Stock Price

Inspiration

We were inspired by the potential predictive power of companies social media data. It's ultimately where the majority of attention of the consumer market belongs so we had the hypothesis it would be a strong predictor.

What it does

We built an augmented ARIMA model that incorporated a stock ARIMA model of stock prices, with the augmentation incorporating social data.

How we built it

We built the model in python, with the underlying dataset using Quandl's Social Media Data, and extensive data cleaning and engineering being a key component of our work.

Challenges we ran into

Definitely getting the data into a clean and manageable data frame to be analysed. We'd always heard about data cleaning being a significant part of data science, but this hackathon definitely re-iterated that sentiment.

Accomplishments that we're proud of

Proud to have participated and produced something that has reasonable predicitve power, based largely on classical statistics and in such a short space of time.

What's next for Team Warwick

Plenty to learn! It was our first hackathon, and hopefully won't be our last!

Built With

• arima
• python
• quandl
• statsmodels