Different kind of data you can analyse with BI or machine learning
I’ve worked with many kinds of data, from clinical data to logistic and supply chain data passing through pharmaceutical data.
Each of them is different and should be treated in different manners.
Let’s what we can learn with each of them.
Clinical data
This is the most difficult kind of data you can encounter during your working life. This kind of data should be treated with caution and attention. It’s not easy to interpret this kind of data due to its high differentiation between people.
For example, i’ve studied for a while data from patient with metabolic syndrome, and each of them had different traits of the same illness.
With a good database construction and a LOT of python it’s easy to find some correlation between two traits of the illness (Weight and Age or Environment and Population trait etc…). It’s really interesting that this kind of data is such delicate and important but at the same time they’re really flexible to any kind of transformation but obviously you need to have a biological background with a degree in biology or bioinformatic.
Moreover, nowadays does not exists a data format used to transfer this kind of data easily from hospital to hospital (assuming that there’s not a national “clinical database”).
This kind of data and all of its implication needs to be studied and observed with a delicate perspective. For example, if my data is stolen or could be seen by my boss, i really dont’ want that my problems influence a possible job hiring or relatioship with colleagues. See? Now you don’t want this too.
This kind of problems needs to be discussed more deeply in another post.
Pharmaceutical data
I don’t really like this kind of data, maybe because they’re less clinical but more statistical than others.
I’ve worked with this kind of data for many times and they’re always the same. Nope you’re not accessing clinical trial data (explained before) but you’re only accessing of the drugs are sold across the countries and which of them are sold more than others. How a pharmaceutical company could improve its business by competing with other companies.
You will see the performance of the Pharmaceutical agent that will rum between private doctors convincing them to sell a specific drug instead of an another. You will calculate of the company performs and how well the sales person is performing across its zone. This kind of data has a little flexibility , just because you’ll receiving the data “as-is” and you will manipulate it by little. The only positive thing is that you can create really beautiful charts with the same data.
Supply chain data
This is the kind of data that machine learning, deep learning and data warehousing (SQL/NoSQL) will be really useful with a lot of potential.
When i first started using this kind of data i haven’t seen its potential but now it’s HUGE.
- Here you will have the data of what a population buy online (with the respect of privacy where you cannot know who is she or he)
- GPS data tracking the parcel where is sent and the road taken from the driver optimizing time an consumptions.
- Learn a lot of integration techniques because you need to integrate your system from the customer ones.
- A lot of machine learning for predicting what a customer would buy or where a driver should go with the parcels optimizing the time.
- NoSQL and SQL databases for different kind of data elaboration and storage.
- Insights for you company and for customers learning patterns and how to compete with others (Yeah just like the pharma but this kind of analysis is done by everyone).
And lot more…
Those are only a little part of this enormous world full of data.
