More Projects

This project was a quick visualization on a dataset that I found through a weekly community known as Tiny Tuesday. Though the data was published over a year ago, it’s still a great dataset to practice some visualizations on. For this dataset I chose to use Tableau again!

The dataset today contains data from the Olympics between 1896 and 2016. The main story were looking at here is how has representation in the Olympics has changed. 

The first visual here takes a look at representation across the Olympic games, what a massive change over the last few years! It’s also important to note, many of the “dips” you’re seeing across the graph is Winter Games vs. Summer Games, as the Summer Olympics is a much bigger event. Also seems to show less representation within the winter Olympics.



![](/images/sex%20by%20year.png)

Next, it’d be interesting to see the top countries by both Sex representation. All 5 top countries by count of male participants, also make up the top 5 by count of female participants. This is fairly logical.



![](/images/sex%20by%20country.png)

Lastly, lets look at the top male and female sports by medal count. Interestingly, Swimming, Athletics and Rowing make up the top 3 for both Sex’s. 

 ![](/images/top%20sports.png)

Please interact with the dashboard yourself, available [here.](https://public.tableau.com/views/olympics-vis/RepresentationintheOlympics?:language=en-US\&publish=yes&:display_count=n&:origin=viz_share_link)
Thanks for reading!




Representation in the Olympics

Interactive DowJones Dashboard

Hi! This post will go over a project I completed to analyze a portfolio of stocks using python and various packages. For the purposes of this write up, I’ll be choosing a few stocks to create a portfolio.

For this analysis, were going to be creating portfolios of the following tickers: GSPC, AAPL, META, MFST, NFLX, GOOGL.

Before we do that, lets look at how these stocks are correlated. We can do this via a heatmap. As expected, stock prices among this group appear to be fairly highly correlated
![](/images/heatmap.png)

This comes as no surprise if you were at all familiar with these companies. This is certainly not the way to build a well-balanced portfolio, but this is for fun so why not!

Now let’s start creating some portfolios from this collection. To do this, we’ll generate semi-random weightings among the stocks, allowing for shorting (negative % holdings) but ensuring the total portfolio has a 100% allocation to the market. For this example, we’ve taken 10,000 samples.

The most likely question after this, how did the portfolios perform? For this, lets take a look at the annualized returns distribution for our portfolios.
![](/images/annual%20returns.png)

Alright, following that it might be interesting to see how our worst and best performing portfolios. The portfolio with the lowest return boasted a -1,358% loss with an annualize std. of 6,528% and an annualize Sharpe Ratio
of -0.208. The allocation of the portfolio was -7171% GSPC, 2152% AAPL, -3286% META.,-19,561% MSFT, 19,185% NFLX, 8,682% GOOGL. On the other side of things, our best performing portfolio boasted 4,834% annual returns with a standard
deviation of 15,863% and an annualize Sharpe Ratio of 0.304. The allocation of the portfolio was 24,384% GSPC, 59,757% AAPL, -46,927% META.,-55,220% MSFT, 76% NFLX, 17,929% GOOGL.

These numbers are are obviously not realistic, nor a feasible investment strategy so let’s look at one that might be. Let’s analyze the Tangency Portfolio, i.e., the portfolio with the highest Sharpe Ratio. This portfolio returns of 83% on annualized basis with 50% std and a Sharpe Ratio of 1.66. The portfolio had the following weightings: -130% GSPC, 130% AAPL, -70% META., 105% MSFT,  1% NFLX, 68% GOOGL. These seem a little more realistic.

Here's what a plot of all our sample portfolios.

![](/images/efficient%20frontier.png)

Lastly, lets dive just a little deeper into our strategy with the highest Sharpe Ratio and see if it would have performed well over time. Here’s a quick distribution of our strategy’s daily returns.

![](/images/portfoliodaily.png)

Next, let’s take a look at it’s rolling 6month Beta to the S\&P 500.

![](/images/portfoliobeta.png)
Lastly, lets look at our strategy’s rolling 6month Sharpe Ratio.

![](/images/rolling_sharpe.png)
Yeah…maybe generating random numbers in python isn’t the way to create a successful portfolio? Even though I might not have stumbled upon the most optimized portfolio, it was still a good exercise! Thanks for reading and be
sure to check out other articles below!


Portfolio Optimization in Python

This project aimed to explore some data analysis within Tableau.
Having used the software a few times prior, I decided to explore some data that
I found on Kaggle that showed a few statistics for Twitch Streamers. While this
was published in 2019 and doesn’t reflect necessarily reflect current creators
on the platform, it was still a fun exercise. The following few paragraphs will
walk through basic analysis and my choice for each figure generated in the dashboard/infographic.

The first and likely the most direct measure of success for
a creator on the platform would be their total number of followers. As we can
see, at the time this snapshot was taken, Tfue was well in the lead with nearly
9 Million followers, and shroud close behind at nearly 8million himself. One
large take away from this figure was the large drop off in follower counts as
we move down the list. Certainly, many streamers had already developed a significant
following on the platform by this time.

![](/images/most%20followed.png)

The next figure, and likely a question that would very
nicely follow the first graphic, who has gained the most followers over the
last year? It’s clear in this figure, while Tfue and shroud were leading in
total followers they faced competition from exponentially growing creators like
auronplay and TheGrefg who amassed 3.9million and 3.5million followers respectively
that year alone.

![](/images/followers%20gained-6fb28ef5.png)

Another important metric for these streamers is the average
number of viewers they bring in. As we can see by the graph, previously mentioned
auronplay was averaging over 50k viewers per stream. This compared to TheGrefg
who had average around 30k seemed to put in him quite the dominant spot. For additional
information I also included the peak viewer count which would represent the
most viewers a streamer as ever had at one time. These large peak viewer counts
would largely be confined to big events, such as the twitch channel run by RIOT
games (creator of League of Legends and Valorant) who had a peak viewership of
639k viewers.

![](/images/avgviewers.png)

Lastly, we look at total Watch Time via a heat map. Interestingly,
or maybe unsurprisingly, the most popular channels or highest viewer average
didn’t make up the top Watch Time minutes. This is likely a result of these
streamers having been around on the platform longer. The top competitors here
were xQcOW and summit1g, amassing 6.2billion and 6.1billion watch time minutes. This
translates to about 10million hours.

![](/images/watchtime.png)

If you’d like to interact with the graphic yourself, please head
over to my public tableau page. This infographic is available [here](https://public.tableau.com/app/profile/thomas5549/viz/TwitchDataAnalysis/Dashboard#1)

Thanks for reading!

 


Projects

Representation in the Olympics

Interactive DowJones Dashboard

Portfolio Optimization in Python

Twitch Streamers Tableau Visualization