Leveraging What We Know About Meetups Part 1
Photo Credit: https://kryptomoney.com/blockchain-meetup-gurgaon-october-6-2017/
INTRO
I admit, when I first heard about Meetup.com about a year ago, I thought the whole idea was kind of strange. It sounded like a dating site. But with a growing interest in Data Science, a previous co-worker brought me to a meetup held by New York Open Statistical Programming. Hesitant at first, I ended up seeing the genius of it all. In case you haven't heard already, Meetup.com is a convenient way to meet others who share the same interests and hobbies as you do. In my case, I saw how easy it was to learn from so many great minds at these gatherings. I knew from that day that these meetups will be vital for my continued growth as a Data Scientist.
One day I saw Meetup.com data posted on Kaggle by Sumit Kumar. Curiosity got a hold of me and I really wanted to know what kind of meetups other people were forming. Was I missing out on something really cool? Then it dawned on me. For whatever reasons people are forming meetups, the topic must be really important to them if they are actively building a community around it. In a way, this site is basically an encyclopedia of what the people demand. And if we know what the people demand, we know how to supply them as well. This blog is a rough sketch of what applications could be built by leveraging what we know about people's meetups.
EXPLORING THE DATA: R SHINY APP
Instead of constantly retyping out code, messing with filters and such, building an app using R Shiny would let me explore the Meetup data with ease. The link to my app is included here.
The dashboard consists of three different tabs you can click at the top:
- Analysis Overview (a quick look at the data)
- Find a Group (locating popular groups within your city)
- Find a Venue (locating popular venues within your city)
CAVEATS
As you can imagine, a data set size containing info of ALL the meetups since site launch would be quite large. Sumit Kumar who posted the data on Kaggle, limited the data set to only include:
- Data from three cities: New York, Los Angeles and Chicago
- Meetup events under the 3 categories:
- Arts & Culture
- Book Clubs
- Career & Business
I was a bit disappointed about the limited categories that I could analyze. But I figured if I could build a template for now, it can be utilized later for when I pull my own data in the future.
TAB 1: ANALYSIS OVERVIEW
Setting the time frame from January 1st 2010 to October 31st 2017, I wanted to see the overall trend of these meetups.
The line graph on the left shows the growth in the number of members joining Meetup groups. Members tend to join more groups in the 3rd quarter of the year, followed by a dip in the fourth quarter. The trend then jumps back up at the start of the new year. This might make for a great Time Series analysis project in the future!
The bar chart on the right shows us that Meetup.com is mainly being utilized to bring Tech communities together. This is followed by Career & Business and then Socializing communities.
With the bubble chart above, I wanted to see if there were any characteristics that made certain events more popular. Was it the category the event fell under? member size? how long the event was? The bubble chart can visualize this all. Interestingly, even though Career & Business events clearly dominate in terms of number of events, it seems Arts & Culture meetups tend to have the most attendees (indicated by the size of the bubble). As a result, Arts & Culture events have a much higher RSVP count (indicated by the height of the bubble). Meetups that last around 3 hours long also seem to be just the right amount for a good head count. Any more and attendees seem to drop. I mean, we've all got plans right?
Tab 2: GROUP RECOMMENDER
Those who are new to the scene may want a quick way to find a group to join. In this case, your criteria may just be:
- What group falls under category XXX?
- What group is located in city YYY?
- What group is the most popular?
If this is true, than the second tab of my app can help! Simply filter out what city you're in and what category is of interest and the chart will list the groups with the most members. For example, I live in NYC and I love games, perhaps I'll check out Playcrafting!
Tab 3: VENUE RECOMMENDER
On the other hand, what if you want to HOST a group? Where should you host it? For an ease of mind, one might want to host their event at a venue that has held similar events. It would seal the deal if that venue had a lot of positive ratings as well. In such a scenario, the third tab of this app would be useful.
If I wanted to host a Book Club event in NYC, simply filter out the city and category, and the top venues will pop up. The Ukranian East Village Restaurant on 2nd Ave seems to be a good choice. AND lots of ratings...huh.
MOVING FORWARD
There are so many things I wish I could do more of with this project. That Time Series Analysis on when people are likely to look and join groups. Another would be to add a feature on the Venue Locator where a user can input an address. The app would then provide the distance required to reach a particular venue. But most of all, I'd love to scrape my own data off Meetup.com and get a closer look into the Tech community...perhaps I will!
About Author
