Conoco Phillips Challenge
Our Insparation.
We were inspired to combine front end web development and meaniful real world data.
What we did.
Our data is analysed by using K-means algorithems, We are able to look at clusters and use heatmapping to draw comparisons using Attributes such as Average Gas prices and GDP.
How we did it.
To construct the data we used a common Data Science tool in Jupyter Notebook. Our goal was to have real world added and for it to be reflecting into a web site. For the Website we are using Express framework, with heroko for deployment.
What was our buggest challenge.
We can now fully respect most data to draw meaniful business solutions requires a lot of research and resources. Our first initial approach was to do machine modeling, but our models were providing meaniful solution. The issue we had was that we did not have actual supervised data with a label. So our random forest, Nb, SVC models were not really working even when we created the labels. Secondly in our clustering models at times we were get our models to not correlate. So we had to drop columns to make the data work. Eventually we were able to draw some conclusions in our data.
What are most happy with.
We are really happy that we were able work on a project from scratch and make on the fly decisions to drop a plan or proceed through it.
What we learned.
We learn that data is super important especially the correct data. Our team Understood that ours of research in collectiong, gathering, analysing, scrubbing data are all important in business decision making policy.
What's next
Our next goal is to gather more data, we can look at monthly data or even daily data and how it impacts different areas in the market. We would like for data to be inserted into the algorithems and the model to updata accordinly with out having to deploy the front end everytime.
Log in or sign up for Devpost to join the conversation.