2020 isn't a 'do-over' year, it's a year for change. And we set out to make that happen with data science! Our CEO of RevGen consulting challenged us to come up with ideas for social or economic change that he could take to the Denver Chamber of Commerce. Simultaneously, Denver announced the 'safer at home' policy, which allowed for restaurants to open their patios for outdoor distance seating. What if we closed the streets for a "block party" feel to give the restaurants even MORE room to expand their seating? Denver loved the idea! And they needed insights to make it happen.

What it does

We built an interactive application that government officials could use to analyze the costs & benefits attached to closing the street to allow for more room for social-distance seating. The decision-makers can select a region or a specific street and analyze the associated costs (how does it impact traffic?) and benefits (how does it stimulate business owners and bring back more jobs?) in order to decide which streets to bring this block party concept.

How we built it

We chose Databricks as our analysis tool to take advantage of our team's diverse skill set and mix Python, R and SQL. We also wanted to use delta tables and various graphing libraries like folium and plotly as well as native Databricks charting capabilities. The first steps included acquiring & cleansing datasets. For example, we performed data type conversions, filled in blanks where data was missing and got rid of duplicates. We also created a new feature called 'business type' that we will use further in our analysis. We built an interactive map using folium packages that shows restaurants and other businesses along with traffic in the potential closure area. Next step is the cost benefit calculations. We built a reference data set with dollar amount associated to the benefits and costs based on manual research. For example: how can we attribute cost to traffic (how many cents per minute is a person's time valued at?). Once everything is translated into dollar amounts, we can compare the costs and benefits to find a net value attached to every street. We also built a delta table to enable time travelling mechanism and run our analysis for multiple scenarios. Finally, we wanted to use data visualization techniques and an interactive application to communicate our analysis to the city officials.

Challenges we ran into

It was challenging to find a data set that was up to date with the Covid environment and any changes that have come along with it. Our Cost Benefit Analysis was derived from traffic data and we know the "safer at home" policies have impacted traffic patterns drastically! We found that city of Denver traffic sets were the most comprehensive, and as a next step we could run our own analysis to see the decrease in cars downtown for this season (anything from standing on a corner with a counter, to image recognition from traffic cameras.)

Accomplishments that we're proud of

We knew this analysis had to move quickly to align with Denver's new 'safer at home' policies. We built our team with diverse skill sets to tackle this problem, and now have a usable POC that our CEO has proposed to the mayor's task force.

What we learned

We learned that it's important to start with the business need, and focus on building an tool that tells a good story and makes it easy to make decisions. We learned that this Cost Benefit Analysis may be biased toward higher priced locations in Denver due to their high margins, so next steps would be to adjust the data set (as needed) to account for mom & pop shops and make sure businesses have equal opportunity. And last, the employee benefit was bigger than we expected! A benefit tied to increasing jobs showed a 2x benefit in comparison to the business revenue.

What's next for Taking it to the Street

We are talking with the mayor's task force around the block party concept. We want to focus on Larimer, Pearl and downtown Arvada first, then we plan to keep pushing this tool forward to extend beyond Denver to all of Colorado and share findings with folks in State Government Office!

Built With

Share this project: