Flat6Labs Industry Mapping
Author: Oliver Tsappis, otsappis@googlemail.com
Written for the non-technical reader.
Introduction

Every startup invested in by the Flat6Labs fund impacts industries, people and markets around the world.
How can they measure these things to improve their business and help the startups in their portfolio?
This project creates a visual map of startups, clustered by their industries to provide insights that could help better selection and the program itself.

Key take-aways
On the insights:

This is simply an exploratory project that actually raises more questions than it answers

On the quality of the model:
Background
What is Flat6Labs?


Flat6Labs is a leading startup accelerator operating in the MENA region, founded by Sawari Ventures in Cairo in partnership with


Their aim is to push innovation in the Middle East & North Africa by supporting early stage startups and entrepreneurs with 3 tools:

Angel investment.
Acceleration program: If selected, local entrepreneurs are trained and connected to a network of experts and investors to improve performance of their startups.
Follow-on investment program.



Their objective is 2-fold:

Give local entrepreneurs the platform to bring innovation to market for the greater good of ideas and economies in the MENA region.
Improve value of their angel & follow-on investment portfolios.



In order to create the largest impact and return possible, Flat6Labs must optimise their startup selection procedure and acceleration program.


To do this effectively, they must understand the complex world in which they operate: industries, people, markets etc.


This is especially important information for the selection committee and program managers. These people need to know the future potential of applicants and in what way to help startups already in the acceleration program.


Being able to model, visualise and communicate these things would allow Flat6Labs to understand the entreprenurial environment better and find opportunities to optimise their business:

What types of problems are regional entrepreneurs trying to solve?
What types of businesses are investors most excited about?
What strengths, patterns or gaps are their in the investment strategy of Flat6Labs?
Where can technology and recent innovation can be applied elsewhere, especially tech developed inside the Flat6Labs portfolio?
What business models are having the largest impact in terms of investment, revenue, employment etc.?



One company that is already modeling this kind of data and is answering these types of questions for their clients is Quid, you can read more on how they do it here (non-technical): https://goo.gl/YeD1EH
The scope of this kind of research can be huge, but we will start by looking at industries the startups are currently operating in and dive a bit deeper from there.
Objective

To visualise the map of Flat6Labs' investment portfolio of startups in terms of their industries.
Imply some answers to the questions:

What types of problems are regional entrepreneurs trying to solve?
What patterns or gaps are their in the investment strategy of Flat6Labs?
Are there any potential market gaps the selection team should look into?



Data

We have available a list of Flat6Labs startups, each tagged with their industries.
I drew up the tags myself
Not all the startups are still currently operating, so the list shows the ideas that Flat6Labs have directly invested in
csv format.
Columns: name | industry.
Each startup can have multiple entries.

Approach

We will create 3 maps (visualisations) in total:

Level 1: This first map will show all Flat6Labs startups in 3 main clusters as an intro.
Level 2: The second map will break-down these 3 clusters into 8 clusters to give more detail.
Level 2 3D: Finally we will can turn the more detailed map 3D to see things in a different light.


In the end we can see how well the maps help us to answer the questions we set out in the objective.

Technical:

Using a k-means method, frequently used for customer segmentation projects, we will be clustering startups based on common tags.
data prep will be all about transforming the data into a sparse matrix, with the industry tags as columns, and the values as binary 1 or 0 to indicate whether a startup operates in it or not.
We can finally map the clusters of startups using Principle Component Analysis (which gives us our X, Y and Z coordinates) to pick out any interesting patterns.

Kyso beta

test

otsapp

Flat6Labs industry map

otsapp

otsapp

Most used tags