My journey into Data Science World

During the initial days of my IT career (early 2000) was always into developing front end and back end applications using Java and related technologies. Writing queries, stored procedures on the database and displaying them on the web page or standalone applications. These were my early touchpoints with "Data" part of "Data Science".

Initial Motivation

However the urge to understand the science behind data started in 2015. This was the time when I had completed my deputation of setting up first ODC outside of India for Bosch at Vietnam. Upon my return was assigned to take up the Program Manager role for BOSCH Safety division in India.

When we talk of Safety am referring to Active safety (Brake Assist, Special Stability Program, Standstill and Speed control) ; Passive Safety (Air Bags) and Drive Assistance (RADAR, Cruise Control Etc.). Had to come up with Roadmap to drive the simulation activities across these 3 units. There are lot of smaller components in each of these safety functions and we had to simulate each of these minor components. One could easily imagine the varieties of inputs and outputs from each of these components i.e. various types of input files (file formats / extensions) and the varied sizes too. Also important were the locations (storage) of these files. Normally there are multiple tools together grouped as "tool chain" for any kind of analysis, forecasts and behavior of these components.

During this time met one of my friend who was working in the data storage in a MNC, introduced me to the world of "Big Data" and "Data Science". I could easily map this to my current assignment and initially went through some technical books (Oreilly publications) on the same. Applied my limited knowledge in initial analysis of the components involved. However had to take up a cross functional assignment in May-2016, handling the Intellectual Property division at Bosch India. Mainly was involved in developing Patent Analytics case studies, building and maintaining the IP Portfolio. It immediately struck to me that this was another case of Big Data and realized that "Data Science" is can be applied in various functions to improvise results. This is when I underwent "Post Graduate Course in Business Analytics and Business Intelligence" course from Great Lakes Institute in collaboration with "The University of Texas at Austin". Most of my batchmates in this course were junior budding data scientists

Implementing sample use cases , Derive Results, Realize Benefits

Case A : Best reinforcement of knowledge is when you apply it to derive insights and benefits. We had tough target to achieve 1 patent per working day i.e. increase the IP Portfolio by >50%. Had approximately 8 GB of data to analyze spread over previous 5 years and cutting across 10 divisions of BOSCH. First I applied the Decision tree to find out initial drill down and later combined it with Lasso Regression to derive at average duration for each phase of patent filing.

Used Python on Jupyter lab environment and also cross checked the results in R (R Studio). R is easier for people with little programming knowledge but have depth in Statistics / Data scientists. Some of the insights were like new age technologies took longer time for evaluation from experts while proven technologies domain took shorter reverts. Also some of the Patent Attorneys had better productivity than others. With all this insights, a task force was formed and implemented the actionable. This resulted in achieving our goal of "1 patent per working day" target and got "Hall of Fame" award from BOSCH India management under "Innovations" category.

Case B: This convinced my belief further that having "Data driven mindset" is the key for predictions, forecasts and translate actionable to derive business value/results. Currently am responsible for Engineering Services and Product operations for Connected Mobility at Bosch India. From day 1 have been looking for use cases which can translate to business value to customer / organization and thereby increasing revenue / optimizing costs respectively. Some of the customer centric use cases are like predicting service maintenance, failures of components while organization specific cases are like automated classification of tickets, optimizing the infrastructure etc. We achieved 99% accuracy for service maintenance prediction use-case while we could predict failure in components with 90% accuracy.

Visualization of data is equally important while analyzing large data and have been using "Tableau" tool at BOSCH for the same. Have intermediate level knowledge of Tableau and have built various dashboards for better representation of data.

Stay updated

Additionally we need to keep ourselves updated with what's happening around Data Science as technologies are changing fast. It's always good to subscribe to Data science websites, read blogs and try out newer software packages and IDEs around Python / R. Additionally am part of intra organization forums like "Tableau Workgroup" and "AIoT Expert Group". Recently have installed "Juno" on my personal iPad Pro to analyze smaller size data like for ex: to analyze my sleeping patterns :-) from my "Pillow" app on my watch.

Closing Remarks

Data and the science in it, has become a part and parcel of my life and feel that journey has just started and a long way to go and am enjoying every bit of it. My suggestion to those who want to start this journey is to look for those initial motivation ->> implement on sample use cases ->> look for results ->> realize the benefits. Once you do these you'll fall in love with this topic and your journey begins.

Wish you all a very happy Data Science journey!!!