Inspiration
Our inspiration was derived from our shared passion in utilising data exploration to generate insightful knowledge that can reveal certain trends, patterns and business strategies. Over the few days, we wanted to explore the reason where the report figures of Sales (Domestic Ultimate Total) and Sales (Global Ultimate Total) differ. With the knowledge that these reported figures can be an indication of the different ways that entities would report their revenue, we can attempt to use our predictive models based on it.
What it does
Our project aims to predict domestic sales of companies in Singapore using various input data from the dataset we acquired.
How we built it
We first explore the data and carried out feature engineering and selection to get our desired columns and data that we think will best fit into our machine learning models. We then proceeded to model the data using various ML models such as Random Forest Classifiers and Regressors, Histogram Gradient Boosting Regressors and Ridge Regression. Finally, we communicated and visualized our results using accuracy scores, adjusted r^2 score as well as diagrams such as heatmaps and line graphs.
Challenges we ran into
Without much prior knowledge of any business jargons, our team had to decipher many of the terms given in the definitions. Due to the intricacies of business-specific terms, it took us more effort to form a connection between our dataset and the complex terms in the business context. Furthermore, we decided to explore the different types of regression models which concluded to have low accuracies.
Accomplishments that we're proud of
Our team managed to figure out that we had different entities reporting their Sales (Domestic Ultimate Total) and Sales (Global Ultimate Total) differently. Upon clarification, we realised that the way that each company report their sales are similar to the way that they report their employee figures. Therefore, we decided to refine our classification model that will fit into our discovery.
What we learned
In our journey, we discovered the essential elements that shape the art of data storytelling. We found that crafting a narrative from our dataset takes collaborative effort and effective communication. Through each iteration and numerous refinements, we recognized the importance of conducting our own research, a proactive step taken to bridge gaps in our understanding of the business context.
What's next for We are the Champions
We would like to discover how our model can be used in real-time data analysis, possibly collaborate with businesses and implement our model while improving it.
