Data Preparation Q3 Sumalika

From Visual Analytics and Applications
Jump to navigation Jump to search

Data Pre-Processing:


Dataset: Meteorological Data
Tools & Techniques:
1. JMP
2. Tableau
3. Excel


1. Check for Missing Values in the given data set:
The Meteorological data set given by the weather department consists of few missing values which can be excluded from the analysis. The two rows that contained missing values were discovered using JMP a screenshot of it is provided.
100%

2. Check for Duplicate Values:
Since there is no unique identifier, no issue if duplicate values.
The date field has 1 value for each, hence no repetitions in date value.

3. Analyse variable distributions:

Date: Refers to date fields for Months April, August and December

100%


Wind Direction: Wind direction ranges from o.1 to 359.1 with a mean of 236 degree.

100%


Wind Speed: Wind speed ranges from 0.1 to 6.8 in terms of m/s units.

100%


4. Data Transformation:


- For Wind Speed Trend Analysis: Join Sensor Data with Meteorological data to obtain the trends in wind speed with respect to chemicals.
Step 1: For each date point available in weather data, identify the average wind speed using Pivot Tables.
Step 2: For each date point available in sensor data, identify the sum of reading for the 4 chemicals using Pivot and map it to its respective dates.
Step 3: The data source (final) for wind storm trend analysis is formed by joining both the transformed tables in Step 1 and Step 2. The table is as follows:
100%


- For Cox Comb Plot: Join the Sensor Data with Meteorological data to obtain the Cox Comb Plot which will aid in identifying the factories with high chemical release while taking into consideration the wind speed and its direction.
Step 1: Convert the wind speed from m/s to MPH as it is given that the given nature park map is divided is 200 X 200 dimensions ( / 12 each) .
Hence convert the wind speed value using the formula:
Wind speed (MPH) = Wind Speed (m/s) * 2.23
Wind Speed (Grids) = Wind Speed (MPH) * (200/12)
Step 2: Map the sensor to their location and name them X and Y. Refer to the table below for clarity. The location column is for information purpose provided by VAST and it is used to obtain X and Y values.

100%


Step 3: Map the meteorological data with the location data. Data needs to be transformed in such a way that all the days and weather data are mapped to its respective sensors and chemical. The transformed data is as follows.
Step 4: To compute the direction of wind on tableau, cox comb plot works on the concept of a plume model.

100%


Step 5: Give path ID's as 1, 2 , 3 to calculate the other co-ordinates of the triangle shown above.

The transformed table is as follows:


70%


The application of the above formula on out table variables:

100%

Step 6: The above transformed data is used to make cox comb plot on tableau ( refer to the Visualization tab under Factories section)