The Ludic II: improving the social platform with statistics

By Paul van Beek, Willem Horsten and Joost Liebregts

Introduction

In January 2015, Industrial Design will move to a new building. Part of this move is a new bar for Lucid, Industrial Design’s study association. The Ludic, Lucid’s bar is now situated in the basement of the main building and will move to a bigger, more awesome room in the Laplace building.

This new start allows for the introduction of a new approach for the bar. As a group we are interested in finding areas with potential for improvement within the statistics. The Ludic has a database containing data such as order size, transaction time, user id and what was ordered. This data was collected over a span of one and a half years and it contains about 15.000 records.

Design goal

The Ludic is a bar, which in its nature is a social platform. Our goal for The Ludic is to improve this platform and stimulate social behavior.

Assumption

Further on, models are shown of average order size per generation, which is a term used for the year in which students start. Student generations are hereafter called by their S-numbers; S08 is 2008, S11 is 2011, and so forth.

From our experience behind the bar, we see there is a correlation between order size and social behavior. This can be explained as, on average, a person does not order three drinks for himself, but he orders those to share them with others.

Explorations

First we looked at the distribution of order-sizes for each generation in The Ludic database. The generations in the database range from 2002 to 2014.

From the database we extracted the size of each transaction for every generation. This data was filtered: we excluded Token Drinks and certain people whom we know do not belong to the regular users of The Ludic. This data is stored in a .csv file and read by Illmo. We only focused on transactions that were recorded between 16:00 and 19:30, which is the timeslot in which the regular weekly Ludic drinks happen. The generations are entered as different stimuli in Illmo.

We tried different models. The model with the lowest LLC (291.19) was a Gaussian model (power=2) with a BoxCox (overall power of -0,0756). Below are the models for each generation. These show the distribution of probabilities for each order size.

modellen van alle jaarlagen (highlight is 2014)

We see that all the models look mostly the same, only varying in peak height. The generation 2004 shows a significantly different distribution, which is much flatter and oriented to the right. Investigating the data shows that we have only very few transactions from this generation. This means that we cannot trust the model for this generation (this also counts for the years 2002, 2003 and 2014). We see the same lack of accuracy in the Thurstone averages of these generations. This is shown in the size of the confidence-intervals.

thurstone average van alle jaarlagen (data values)

Since the older generations show a lack of accuracy, we also looked at the differences between each generation that is still officially present at Industrial Design (2008-2014). We eliminated the data of the older generations and recalculated differences. We took the year 2008 as a reference, because on average it is the year with the largest orders.

Thurstone Difference for 2008 to 2014

Even though we cannot draw conclusions from these graphs, we can see some sort of trend. Younger students (on average) seem to order smaller size orders than older students. This corresponds with our personal observations in The Ludic.

National & International students

The database also has information about international and national students. We found this an interesting direction to investigate in the context of the social platform. Using Illmo we created two new distributions of order size: one for international and one for national students. It should be noted that this data for international students only counts for those who were registered with the Lucid study association. Below the models and averages are shown. The model in orange are national students (stimulus number 2).

modellen van NL en EN (highlight is EN) thurstone average van EN(2) en NL

 

Even though we have little data of international students we can see a significant difference in average order size. Looking at the distribution model we also see that national students still order either

Design concept

To reach our design goal we need to improve the social interaction between students in the bar. Therefore, based on our assumption that order size is related to social interaction, we need to widen the distribution of order sizes, and thus increase the average order size.

We developed a system that confronts the students in The Ludic with their social behavior, as the average order size is displayed on a big screen. By this intervention we wanted to test if the social interaction data could be improved by showing that data and adding a competitive element. We chose to show the average order size as a simplified version of the distribution, because it is more understandable for the students.

prototypeconcept

After testing this system during one drink in The Ludic, the results were clear. Almost all generations had a bigger average order size during the drink than over the past one and a half year. Also students of one generation united and ordered together to get a higher average score. Based on observations can be concluded that the social interaction between the students of a generation increased significantly.

 

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s