We were inspired by the effect that social media has had on the marketing techniques and reach of the high-end fashion industry. Companies like Gucci and the new luxury streetwear brands have been able to interact with younger generations which represent nearly 40% of their market through social media and we wanted to find a direct correlation between these and discover if we could use follower count or social media reach in order to create a metric to predict financial data such as capex, share or bond prices, etc.
Due to a difficulty, we decided to reconsolidate our product, as we had issues with viability. We were inspired by the trends in forex prices and how they are analysed and understood What it does
What it does
The model predicts forex pairs bid average price using a moving linear regression, as the model learns from the previous 25 days
How I built it
Spent a significant amount of time trying to understand what we want from the data, and how we were going to extract the information. We then subdivided tasks among our team for each member to construct
Challenges I ran into
12 hour change of plan Interpreting data
Accomplishments that I'm proud of
We managed to produce a model We worked well as a team Learned to use, keras, scikit-learn, numpy and pandas effectively, as well as understanding the maths and statistics behind the product
What I learned
Python Pandas Tensorflow keras Scikit learn CNN Kernels
What's next for Quandl_ForexPredictions
Minimimze error by using machine learning to try and parametise the start of a downward trend in order to reduce lag which could lead to a loss.