Human Activity Recognition Using Smartphones Data Set : Frame the Problem

Now, let us git our hands dirty with some real data, to practice what we have learned in theory until now. Shall we begin :).

1. Who is our client?

It is Smartlab. 

2. What exactly is they asking us to solve?

The Smartlab has developed a publicly available database of daily human activities that has been recorded using accelerometer and gyroscope data from a waist-mounted Android-OS smartphone (Samsung Galaxy S II). The six studied activities are: Walking, Walking Upstairs, Walking Downstairs, Sitting, Standing and Laying down.

So, they need from us, as data scientists, to build a model that can detect these activities automatically.

3. How can you translate their ambiguous request into a concrete, well-defined problem?

By following data science process of course:

1. Collect available row data from available resources (experiment website, database, images. logs. etc..)
2. Process the data to (remove unneeded data, add more needed measurements, reshape data into exploitable shape, etc ..)
3. Explore the data to define our trends and know what to do next.
4. Build our machine learning model to resolve the problem, clearly this is a classification problem :).

4. Is this data already available?

Lucky us, the data is available here. and this is how they collected it.

Next, we begin our journey to know our data and begin processing it.