# A3-ShimulSachdeva

## Domain

Top viewed shows and channels in the US (in the week Feb 12-18, 2010)

## Question

What factors help a network gain popularity?

## Database Source

Nielson Media Research (http://en-us.nielsen.com/rankings/insights/rankings/television)

## Overview: Top 4 Networks

Here is a pie-chart depicting the share of viewer base. The data pertains to the top 4 networks that air the top 20 shows. (Note: This is data for Feb 12-18 week).

## Hypothesis 1: Show Popularity Helps Gain Network Popularity

### Approach

My first instinct was to look at the most popular shows. I wanted to find out if show popularity is in direct correlation with network popularity, i.e., the network with the most popular shows is also getting the most viewers (in numbers). So I made a visualization (2 iterations) showing what shows are popular (first iteration) and what channels they belong to (second iteration).

### Result

As can be seen, FOX has the most popular shows, namely American Idol (Tue) and American Idol (Wed) and yet CBS has more viewers than FOX. A notable difference between FOX and CBS is the number of shows in top 20 that they air. This led to Hypothesis 2: Number of Popular Shows/Network Affects Network Popularity

## Hypothesis 2: Number of Popular Shows/Network Affects Network Popularity

### Approach

In order to explore the viability of this hypothesis, I decided to chart the number of shows (among the top 20 popular shows) against the network and compare the lengths of the bars. Here is the bar chart:

### Result

The chart explains why CBS (with 9 prime time shows) has more viewers than FOX (with 4 prime time shows) despite FOX having the top rated shows under its belt. But I noticed that both FOX and ABC have 4 prime time shows each and yet, FOX has more viewers than ABC. Hence the hypothesis seemed to work only partially and I wanted to analyze the data further to look for more contributing factors. This led to Hypothesis 3: Popular Show Timings Affect Network Popularity.

## Hypothesis 3: Popular Show Timings Affect Network Popularity

### Approach

In order to explore this claim, I first made a line graph for the shows and their timings. Here it is:

But after making this graph, I found hard to analyze it. First I had to annotate it appropriately because it was hard to identify the points on the graph. In order to understand the dataset better in relation to show timings, I iterated through different visualizations. They are as follows.

#### Iteration 1

This iteration uses a stacked approach to see what networks air what shows at what times. But due to the multiple shows, the data looks slightly confusing. Hence, I decided to go to iteration 2.

#### Iteration 2

Here, the information about the specific shows has been removed and only network aggregates are taken into consideration. This simplifies data a lot. A quick glance shows us that FOX and ABC have shows that are spread over different time frames. FOX only plays its top shows at 8 or 9 PM; ABC on the other hand, has its shows spread over to 8, 9 and 10:30 PM slots. This might explain why, despite having the same number of total shows (among the top 20 rated shows), FOX gets more viewers. Perhaps, 8 PM shows get more audience than 10:30 PM shows.

## Final Visualization

Based on the iterations above, I decided to encode more information in the visualization. By color coding, I showed that CBS gets most viewers and NBC the lowest. Also the 2 pie charts on the right show a quick view of viewer share (#viewers/network) and show share (#shows/network). This visualization helps better to understand why:

• CBS has more viewers (even though FOX has the top 2 rated shows) - possibly since it airs 9 shows per week, while FOX airs 4 shows per week
• FOX has more viewers than ABC (even though they air the same number of shows) - possibly since FOX airs most of its shows at 8 PM, while ABC does not.

While iterating, I changed my questions and hence my visualizations changed accordingly to help me answer those questions. It is interesting to document this progress and view it with the final outcome.

I would like to acknowledge the following 2 shortcomings:

• The data is only for Feb 12-18, which also had some shows not airing due to President's Day Holiday
• The data is skewed towards seasonal shows (like American Idol)

## Future

Some other directions I wanted to explore are:

• Relation between popularity of a network and the age group of its viewers
• Relation between popularity of a network and genre of the shows it airs