Effective steps to avoid Rambo method of machine data cleansing

  • Home
  • Digital
  • Effective steps to avoid Rambo method of machine data cleansing

Data cleansing in a key process in data science. It remove major errors and inconsistencies that are inevitable when multiple sources of data are getting pulled into one dataset. And it map the different functions and what your data is intended to do and where it is coming from your data.

As the name “Rambo” sounds, one thing that suddenly comes to our mind is a fictional character in the Rambo saga. The term “Rambo” is used commonly to describe a person who is reckless, disregards orders, uses violence to solve problems, enters dangerous situations alone, and is exceptionally tough, callous, raw and aggressive. Well, in data cleansing we use exceptionally tough methods to clean our raw data.  Data cleaning is one of those things that everyone does, but no one really talks about. In other words… garbage in gets you garbage out. Obviously, different types of data will require different types of cleaning. Suggesting following steps and guidelines for your effective data cleansing process.

6 step data cleansing process for efficient machine data analytics:

  1. Standardization – Clean the raw data from equipment by placing the right information in the right columns and cleaning all data (including address management)
  2. Data enhancement – Add data that enhances the information. For example, add SIC code to help identify records from same industry;  or match to industry reference data such as DUNS numbers.
  3. Remove duplicate – Plan a series to remove duplicate data values. Clusters of duplicate records are created as potential records to be merged.
  4. Automatic merging – Define merging rules to create a golden record. A  good proportion of the merged records can be automatically be merged. All related entities will also be merged to the golden record.
  5. Manual merging – Same as stage 4, but the decision to merge record is made by a person. Some cluster of information will be very difficult to merge automatically.
  6. Deletion – Finally, removes all those records that were duplicate , but is not golden record.

Above steps are involved in data cleansing. But, the challenge here is generalize the data cleansing. For instance one chiller plant is air cooled and other chiller is water cooled. We need to build an automated method which can do data cleansing in an efficient way. Building automated method is a tedious task. Reason behind this lack of knowledge is to cope up with the modern technology, lack of skilled worker and difficulty in analyzing data.

We at Maxbyte Technologies build automation and standardization for data science process for industrial equipment analytics requirements and make predictive analytics process more efficient. Want to know more? click here.