IIoT Data Analytics

Client's Overview:

A leader in professional cooking equipment design, manufacturing and service, supporting an extensive portfolio of premium branded product lines for commercial kitchens and the needs of today’s commercial food service operators.

Challenges:

Boilerless Food Steamer (Commercial Food Steamer) – deployed as a commercial kitchen product had an inbuilt connected system to gather critical health data but was lacking data analytics capability to derive data-driven insights and address specific business goals.

Solution & Approach:

The need was to study the extensive set of data acquired from the product and build the data analytics application to provide actionable insights.

To build the data analytics system a methodical approach was adopted where understanding & use of the story derived from the acquired data was used to define the goals, covering-

Creating data storage of acquired data

Defining business goal definition in mutual discussions with the client
- Ideation & defining broad goals like improved performance, better operations and more
Exploring & preparing data for analysis to include- data normalization, filtering duplicate or erroneous data, and creating derived data (e.g., time durations, scaling, conversions, etc.)
Exploring the data set by creating visualizations for various trends, co-relations, and dependencies
Creating the ability to gather insights from the data that can be used to feed further steps to meet business goals
- Usage insights- such as usage patterns with respect to time of the day
- Performance insights- such as steamer temperature drops with respect to door status open time
Building algorithms
- Analytical – algorithms based on gathered insights to validate & correct the insights by applying insights to the larger dataset
- Predictive Deployment and Evaluation- algorithms based on outcomes of analytical algorithms which will predict error or usage conditions and warn users
- Control Deployment and Evaluation- algorithms based on outcomes of analytical and predictive algorithms which will correct or control the error conditions automatically

Evaluation and reporting of results of the deployed algorithms to check the achievement of business goals and to further train the algorithms

Results

The developed system provided the ability for analysing multiple data-driven insights to address the set of business goals defined as per the client’s requirements. This included the key business goal “to reduce energy costs” which was addressed by understanding the data and story derived from data covering –

Identification of parameter sets that are dependent & carry relationships
Building visualizations for each set to gather insights and using them to identify and arrive at algorithms such as pattern recognition, that can help in achieving the defined business goal

Outcome:

The developed system was scaled into a comprehensive data analytics solution providing-

Predictive insights – for service and fault pre-emption
Analytical insights – for operational and design feedback
Control insights – identify needs and means for controls
Business insights – possibilities for energy, cost saving and more

Technologies:

Python

Jupyter Notebook 6.3.0)
Visualization libraries- Pandas Visualization, Matplotlib, ggplot, Seaborn, Plotly
Visualization Plots /Graphs – Scatter plots, Bar & stack-bar charts, Box plots, Histograms, Heat maps, Area charts, Correlograms