How to Communicate With Data
Data analysts must be good communicators. There is no debating that. Great insights which data analysts might discover will sink to the bottom of their managers’ desk paper stack, missing out on opportunities just because the presentation of the information was not impactful enough. As a marketeer, you also have to be able to communicate data effectively to your audience. In this case it might even be more important to be able to do that, as your audience has no obligation towards listening to you.
During this post I will showcase how I have dealt with the data analysis and communication process by teaching my friends how investing works. None of them possessed knowledge about finance, so the need for effective communication was essential. To showcase all the information into a digestible form I have opted to create 3 infographics which can be found here:
- How much money do you need to start investing?
- How long should you invest?
- Risk-return balance strategies
I have used Python and Pandas library to process the data and Tableau to visualize it.
Understand your audience’s needs
Before taking any step towards analyzing data behind your desk, you must get out there and figure out what your audience is looking for. This might sound logical, but so many people underestimate the value of this process. They build their project based on assumptions and the outcome is not properly designed for the audience. The tricky part of this, however, is that most of the times the audience could not possibly know what they want.
Which makes this process even more important. Being the expert on the matter, you have to come up with questions that will extract the essential information from your audience. By conducting surveys or interviews with your audience, and having a structured approach you will get to the root cause of your audience’s need. In my case I have found out that the assumptions my audience made about investing are:
· Investing is risky.
· You need to know a lot of math and finance.
· You need a lot of money to start investing. I will do it when I turn 30.
· I would only invest for a short amount of time.
With these insights extracted from my audience, I can now come up with ideas about what information I must showcase.
Gather your data
When you have an idea of what your audience really wants to find out, it’s time to go grab that raw data. Depending on your project, data might be easier or more difficult to find and retrieve. In my case historical price data of several stocks was available with the use of Yfinance library in Python. With the help of this library, data about stocks can be retrieved with a few lines of code:
pip install yfinance
import yfinance as yf #retrieving monthly price data about SPY exchange traded fund
spdr=yf.Ticker('SPY')
spdr_historical = spdr.history(start="2000-06-02", end="2020-06-07", interval="1mo")
Preprocess the data
It is said that a person working with data spends 80% of their time cleaning and preprocessing that data. In most cases raw data will not be neatly displayed. It will be filled with missing values,duplicates, wrong variable types or improperly organized rows and columns. Moreover, the necessary data for analysis may not be there, so it must be created. Here are some steps to deal with this process:
#check na values and forward fill
spdr_historical.isna().sum()
spdr_historical.fillna(method='ffill',inplace=True)#drop irrelevant columns
spdr_historical.drop(['Open','High','Low','Dividends','Stock Splits','Volume'],axis=1,inplace=True)
Choosing how to deal with missing values depends on the kind of data you are analyzing. In some cases, it can be filled in with the forward values (as exemplified using ffill method), backward values (bfill method), mean values, zero or dropped entirely. It is important to deal with values displaying NaN because otherwise it might lead to problems when processing further.
Visualize your data
Now that your data is preprocessed, it is time to visualize it. Before actually using a software tool to visualize it, take a step back and think about what kind of data you are dealing with. In Tableau the data is divided between Measures and Dimensions and further into continuous or discrete. Prototype your graph using pen and paper. This will offer you a perspective on how to properly plot the data and will reduce the amount of work in the software tool. Here are some of the visualizations of my data:
If you need inspiration into how to properly visualize data, I recommend a book called: “Storytelling with Data” by Cole Nussbaumer Knaflic. There you can find the best practices of effectively displaying your insights in order to create impact. Try to be as explicit with your data by leveraging color to highlight or fade information, use the title to give a call-to-action or the conclusion of the visualization and declutter your graph as much as possible. Think of the most efficient way of displaying the necessary data.
Be a storyteller
Effective data communication does not only mean plotting your numbers onto a graph. No one would look at a bar graph and immediately draw out the conclusions. A data visualization needs to be accompanied by a description. People love stories for a reason: they manage to carry the reader through a world of imagination without putting too much stress on his/her brain power. Here are some examples from the infographics I have created:
One thing worth noting here is that the tone you use to communicate the data depends on context. In a business environment the communication must be as efficient and professional as possible and not use figures of speech. In a marketing campaign for a new consumer product, the tone can be more playful. Nonetheless, a good story can get your message across in either environment.
In summary
During this post we have taken a look at the data analysis and data communication process. We have started with field research which will offer us great input for our data analysis. Afterwards, we took a look at the data gathering and preprocessing steps, which allow for modelling the data into a desired tabular shape. This is necessary in order to easily plot your data on a graph and visualize it better. Finally, we focused on the importance of conveying a message out of the data for the audience to better understand our insights.
