Do you often face difficulties when it comes to knowing the kind of information to extract from data? You're not alone. I have experienced this issue severally in my journey to becoming a data analyst.
The truth is, most of the time your biggest challenge isn't creating the charts. It is about getting your information and forming a narrative - an insightful narrative. So how do you begin? The purpose of this article is to enlighten you on how you can choose better narratives for your personal projects. The following steps will aid you as you prepare your mind to handle a data analysis project.
What are your topics of interest concerning the chosen dataset?
If, for instance, you selected a dataset about road US accidents, you need to select your topics. A topic in this context will help you to be more specific in your analysis and not be overwhelmed by the data, especially large datasets. Choose one or more topics that you would like to explore with this data. For example, you could choose "Accidents in the State of Georgia" or "The Role of Weather in US Accidents". The selected topics in this case will be those that you are interested in and would like to explore.
How much data is available to cover your selected topics?
You also want to be sure that the data is enough to cover your topic of interest and answer relevant questions you will be exploring. This is why it is a good idea to have more than one topic of interest. If there is no sufficient information about the state in the case of topic 1, you can easily switch to topic 2. The same goes for topic 2.
Choose a type of analysis
There are three types of analysis; descriptive, diagnostic, predictive, and prescriptive analysis. Here is a summary of these various types:
Descriptive statistics focuses on "what happened?" In this case, you will be delivering insights based on facts drawn from the data. The statistic measures such as mean, median, and IQR are valuable for this type of statistic.
Diagnostic statistics dwell on the why. You want to give information as to why certain things happened. This sometimes is about evaluating the relationship between variables. It could also be about hypothesis testing in other cases. It could even be both depending on the situation. You could for instance form a hypothesis like; " the most severe accidents occur in winter ".
Predictive Analysis involves building machine learning models to predict future outcomes. In the case of accidents, you can predict the severity of accidents for factors like climate, temperature, longitude, etc.
Prescriptive analysis answers "What can be done?". It is about making recommendations based on discoveries. This analysis can be made after carrying out predictive analysis or descriptive analysis or even both depending on the situation.
You can use a combination of these various kinds of analysis if you see the need to. You can also start with one kind and later on continue with another as you get more comfortable with it.
Have it in mind to explain the relevance of your findings
It is relevant to explain how useful your information is. It's not just enough to show the plots. You should also make sure you include how useful this information can be. For instance, in the case of accidents, your analysis can show that certain roads experience severe accidents in the winter season. Go ahead and add this valuable information to your report and state your recommendation(s).
These are all my tips for handling self-guided projects. All the best in your journey. Is there anything you think is necessary to include in this guide? Please let me know in the comment section.