Big Data AchieveMint
I had an opportunity to talk to the CEO of AchieveMint, Mikki Nasch about big data and data analytics approaches. The discussions were inclined more towards the healthcare industry. It was definitely a great learning experience.
Before I start, some background about Achievemint. It is company which provides reward points for tracking personal health activities which is already tracked in application. These applications various from wearable fitness monitors like fitbit to social media postings like in Tweeter, foursquare etc.
The company uses this information to understand various patterns and trends with relation to the healthcare industry and the user in specific.
In this blog I will try to highlight the various learning’s I had through this conversation.
1. Business problem definition must always be the first step. It would be effective to do all the data processing after we know what we are looking for from the data we have.
2. Usefulness of data, when data is analyzed its fine to have data which doesn’t give any signals even that is an essential learning
3. Interesting data may not always mean returns, at times a lot of investment is done on the tools to analysis the data which might provide interesting signals but the returns provided by the signals to the business might be less
4. Raw data is valuable, it is important to store and manage all the raw data. The raw data is generally processed and the processed data is stored separately and utilized. They are possibility that certain parts of the raw data were not important during the previous processing, which might be useful for us in the future. Hence, it is essential to store this raw data.
5. Unsupervised learning model, unsupervised learning model will not work effectively. It is better to go with the heuristic approach where the domain knowledge drives the analysis to the right and effective direction.
6. R&D and real-world coexistence, it is important that R&D and the real world coexist. The results from R&D must be verified time to time with the real world happenings(social media data)
This conversation definitely helped me to understand the real world difficulties of implementing of big data and the challenges an organization would face to gain returns.
