How to kick off your career in data and business analytics?

The power of data

The fuel that drives the entire business world today is the Data. Whether directly or indirectly, every business decision revolves around data. A person who can master in understanding and analyzing data, can have a really great career in coming times. Now comes the question, how can people like us, who are freshers or wish to start their career in data analysis, be a contributor in this.

It is very simple. Every company has huge amount of data. All they need is people who can make sense out of this ocean of information and help them solve their business problems or give their business a boost by making some meaningful data driven decisions.

Let’s learn 10 vitals to get going in the field of Data and Business Intelligence

1. SQL (Structured Query Language) – The first step to big data analytics is to be able to access data. SQL is a necessity to get the required data from the data warehouse and various databases. You should to learn how to pull the needed data, how to do roll ups at required levels and get the stats that you need for your analysis.

People usually think that knowing simple SQL syntax and learning how to join tables would be enough. But things are very different when it comes to actually working on industry level data.

Efficiency is one of the challenges in real projects. There can be two SQL queries that do the same task but the run time varies a lot. It is the way you aggregate your data or hit the source tables.

Look at the example below-

The runtime for Q1> 11hrs and that for Q2 is ~52 mins. It is because Q2 filters the required records first and then joins the tables.

It is always advisable to pick only the required records for your analysis. So always apply the necessary filters.

Another way is to create temporary tables for your analysis. The idea is to hit the database source tables once and pick your filtered data to create a table for your analysis. Then you can do the analysis, roll ups and aggregations on this table to increase the speed.

E.g.

To know the kind of data in the table, always limit it to few records instead of pulling the entire data set.

E.g. Select a, b, c from table1 limit 10;

Such kind of hacks are learnt and developed over time. Each analyst develops his/her best practices while working on the projects.

2. Data cleaning- A lot of the time of data analyst’s job is consumed in cleaning of data. You need to make various kinds of preventive checks that are placed at the entry level of your data. These checks prevent the unwanted data to enter to your system at first place. Always ensure that you put these data quality rules in the initial pipeline of your data.

E.g. These could be like you can’t have a 6-year-old child own a credit card. Similarly, you can’t have age in negative or like 150 years or so. So, these kinds of checks are thought and implemented by a data and business analyst only. Similarly, if your data has null values, missing values, you need to take care of that.

Don’t let your analysis ruin due to bad data

3. Data transformation- The essential step is transforming the data according to your need. E.g. Suppose in a gender column, if you have entries like F, Female, etc. you need to transform them to a single type of entries. Similarly, if there is local timestamp column, you need to convert it into a GMT time zone for consistent analysis. These kinds of activities come under data transformation and after this your dataset is now ready to move into the analysis.

Make the data look like you need.

4. Excel- This is an analytical and reporting tool. Analyst should be comfortable and efficient with excel. Learning shortcuts for excel saves a lot of time. When one starts learning excel, then one realizes that it is an underrated tool but has a lot of capacity to do many complicated tasks in just a few clicks.

Excel visualizations including charts and graphs are an effective way doing analysis and view trends. Pivot table and V & H lookups are super effective tools to get quick statistics from data. Excel macros and VBA are a great add on to learn as they are primarily used for creating automated excel reports and reducing manual intervention by combining lots of steps into just one click.

An example of simple excel report showing look up tables and charts

5. Tableau- This is a data visualization tool used to create user interactive dashboards for displaying all the analysis. Every company creates dashboards for various use cases and different stakeholders. These are one of the effective and easy ways of showcasing the analysis and statistics.

Mostly the descriptive analysis is showcased via dashboards. The intention behind this is, not to repeat any mistakes from the past and to analyze what strategies are working and where is a room for improvement.

Here you can have options for lot of roll ups and aggregations, for various trends and plots, for different time periods and filters, right in front of the user. User can click & apply filters and chose options for what they want to see data.

Dashboards can reflect real time or historical data. Dashboards are crated for leadership and business leaders to know the numbers and performance trends of various sectors within their business. Companies can have their inhouse dashboarding software as well.

A snapshot of a tableau dashboard

6. Python- Python is the language used for analysis that tells how things might look like in coming times. The motive behind predictive analytics is to see and be prepared for what may be coming their way in near future. This can be done by putting the historical data to use, with an idea to see the past behaviour and predict the future trends. This may give leaders actionable insights to grow their business to new horizons. Companies aim to create models that can provide them with ideas to enhance business growth.

E.g. This is how business decisions are made around data. All such models are made via python language mostly. Also, python has good visualizations as well that help to compare trends/ patterns via charts and graphs.

Various types of Data Analytics

7. Machine Learning and Deep Learning- These are the analytical techniques that actually create the models. Suppose an airlines company has the customer feedback comments. Now they want to analyze those and see what can they improve.

As human tendency says, we tend to write comments in feedback form more for negative experiences than positive ones. So, it becomes all the more important to see what is not working for the customers so that it can be fixed.

Now here is where data science technique of Natural Language Processing (NLP) comes into play. The NLP models are designed to know the most painful words used in the comments that give an idea about the thing that is most talked about.

E.g. For an airline, if there are words like stale food, less variety, insects in food etc. are found most frequently, it means that customers are mainly disliking the food and thus company can work on that. So, there are many such use cases of the models that are designed by machine learning and it plays a very important role in many business decisions.

Machine learning put to use for classification of customer feedback comments based on sentiments.

8. Statistics- It is equally important to learn and know the statistics to help you understand the way numbers behave and how are they related to each other. In statistics analysis, things like standard deviation, variance, correlation, skewness of data, various distributions of data, descriptive statistics of data, frequency distribution etc. come into picture. These help in comparison of different data sets and helping you guide on how to proceed with analysis.

E.g. If there is an entry with the value of 10000000, whereas all other entries are in range of 1000’s then this one entry will affect your analysis. It will not give you true picture of the dataset.

If you exclude this one value, then your dataset has fair distribution of numbers and you can do analysis. This kind of cases are discovered by statistics. SAS is a good tool to learn for analysis.

Types of data distributions

9. Putting it on plate- This is by far the most crucial part in analysis. The critical thing is to bridge the gap between the data and the insight. You got to have that strong skill of putting all of your hard work and efforts in the best presentable format in the minimum space and time. It’s the creative art of storytelling.

Your story should have that smooth flow of how you drove from that huge chuck of raw data to cleaning, transforming, analyzing, deep diving and then reaching to the inference and conclusion you are presenting. The main crux should not get diluted in the large set of numbers and aggregations. It should look like an easy transition from problem to solution. That is all it boils down to.

Idea is the take away.

10. Other tools- Spark, Hadoop, R, Power BI are few tools you can learn for advanced analytics.

Apart from technical skills, what else is needed to become an effective data analyst?

1. You need to have the patience to deal with literally big amount of data. You will have to make your hands dirty in the unclean and messy data.
2. You got to be playing with numbers and for that you need to have an absolute love for them. Then only you will be able to discover those hidden things and solve mind -boggling mysteries.
3. You need to have an eye for the data to find those hidden insights that no one could see till now. That is going to be your USP.
4. You need to have the creativity to put all the analysis and findings in the best possible and most effective visualizations that cover your story to the best.
5. Keeping an open mind is very important here. You can’t perform analysis by keeping a pre assumed notion in your mind. Because then you will try to look at the data in a way that convinces you to believe that notion. You won’t be able to see what data is actually saying.

The best way is to challenge your findings/hypothesis so see what else can be discovered. Sometimes, data has a different story to tell that what you expect.

Why data analysis as a career path?

1. It’s an interesting amalgamation of the technical and business aspects along with big data. You can actually create wonders in the combination. It is neither pure coding and technical thing, nor absolutely business related. It’s the kind of perfect trades off between the two.
2. There is a lot of room for career advancement and growth. The career path on data analytics is an approaching its peak. It’s going to be the most demanding field in coming times.
3. The icing on the cake is that the kind of tasks you will perform each day, would lead to significant business impact which you could see right in front of you. This gives a good satisfaction at the end of the day.
4. Data is the open-ended sky that has no limits. You can do endless magic with the same data and still it won’t be enough. So, you need to have an open mind and out of box thinking of what you can do with the same data that wasn’t done earlier.

This is something that can’t be taught in any course or classroom. It can only be learnt as one works and grows over time.

Companies actually need such people who can help them business reach heights. They don’t limit data analysts with the boundary of things. It’s your creativity, imaginations and interpretation of things that has to come out to bring new ideas and solutions to the table. Attention to the details is what makes the difference.

This was a flavour of how the world of data and its wonders look!

ABOUT THE AUTHOR