IBM Watson provides predefined analytics that can be applied on data available at a personal level and in a corporation. A lot of analysis is possible without having to write complex programs required to achieve the same. The data needs to be structured data in a tabular format. The data can be uploaded from Flat Files or from a Excel file,
When the data is uploaded, IBM analytics evaluates the data quality. For the data, it is ideal that the first row (or first line in the flat files) contains the column headers. If this is not the case, Watson can detect the same and ask for resolution.
I used the data I have been tracking for all my expenses to get to know the features of the IBM Watson Analytics. I use this data as I have all the rights on this data and thus do not need providing any clarifications for use in this blog.
When I uploaded this data, it evaluated to 67% quality.
A view of the data. This is a simple piece of data. However, Watson discovered a lot of information from the same. When Watson imports the data, it automatically detects categories if they are available in the data. Notice that the raw data only contained the date of the expense. Watson created 3 more columns as “Year (Date)”, “Month (Date)” and Day (Date)”.
On clicking the data item, Watson provides a few questions that can be asked of the data by itself. One of these can be chosen OR a question in natural language can be asked to Watson.
I started by selecting a predefined Starting Points – What is the breakdown of Amounts by Year and Category?. Watson provided this output.
It also provided a Word Cloud of the Expense Heads.
I could refine the categories to only analyse the expenses, deselecting Savings from Category.
I could create a predictive model for my expenses.
It prepares the decision tree and you can select which part of decision tree to use.
It is possible to create different visualisation. I could prepare a Packed Bubble Chart for all expenses over the different months. Notice that it shows the proportions of the different expenses over the months. This chart includes Savings also.
The same packed bubble chart excluding the Savings.
When there is data from a large enterprise, a lot of analysis would be possible with Watson. Also, these analysis can be shared by email as PDF attachments which can be used a presentations and discussion points.