Resources

Analyzing Data



This section assumes that a number of submissions have been uploaded for the "M and E Form" by the data collectors. If this assumption is correct then the data submitted can now be analyzed using this web app.

In this walkthrough, questions would be asked of the data submitted and the steps taken to get the corresponding answers will be delineated and the graphical results presented.

Note that only the following users can carry out analysis on data submitted:

  • All active Admin users with the View Data role set.

  • tAll active users with the Analyst designation if the form is not assigned to a team.

  • All active users with the Analyst designation that belong to the same team as the form if the form is assigned to a team.

See Manage Users for more information on user privileges, roles and team assignment. Also see Manage Forms for information on restricting form access to members of a specific team. Finally, the Data Analysis page gives directions on how to query form data.



Slide 1


image


Analysis Question: Of the data collected so far how many buildings have defects and how many do not?

(1) On the Dashboard click on Data Analysis -> Aggregates/Chart Based Analysis menu.

(2) Select the M and E Form in the Primary Form dropdown box.

(3) In the Group By Field dropdown box select the Any Defects field.

(4) Since this is a counting question, any field in the Form Fields section can be selected. The Count (Signboard) field is selected in this example.

(5) Click on the Table icon as shown in the image above.

In the query result presented, notice that 8 of the 12 buildings had defects according to the data submitted.

Slide 2


image


This is the same query as that of the last slide but displayed as a bar chart.

Slide 3


image


Analysis Question: What is the frequency distribution of the number of defects in buildings in the data returned?

(1) On the Dashboard click on Data Analysis -> Aggregates/Chart Based Analysis menu.

(2) Select the M and E Form in the Primary Form dropdown box.

(3) In the Group By Field dropdown box select the Any Defects field.

(4) Select the Count (Signboard) or any other field in the Form Field section.

(5) Click on the Filter icon.

(6) In the dialog box presented, select Any Defects in the -Select filter field- dropdown box.

(7) In the -Select filter operator- dropdown box select =.

(8) In the -Select Single Option - dropdown box select Yes.

(9) Click Add Filter and then close the filter dialog box.

(10) Continue on the next slide.

Slide 4


image


Click on the Bar Chart icon to display a result set similar to that shown above. From the query result returned, it can be seen that 3 buildings have 1 defect each, 4 have 2 defects each and 1 has 3 defects. Note that the X-axis represents the number of defects whilst the Y-axis is the number of buildings.

Slide 5


image


Analysis Question: How many of these submissions took more than 10 minutes to complete?

(1) On the Dashboard click on Data Analysis -> Aggregates/Chart Based Analysis menu.

(2) Select the M and E Form in the Primary Form dropdown box.

(3) In the Group By Field dropdown box select the Survey Ack field. This field was chosen because all forms submitted have a similar value for this field.

(4) In the Form Fields section select the Count (Signboard) field or any other field.

(5) Click on the Filter icon.

(6) In the dialog box presented, select Meta Data [timeSpan] in the -Select filter field- dropdown box.

(7) In the - Select Filter Operator - dropdown box select >.

(8) In the DDD:HH:MM:SS textbox enter 000:00:10:00.

(9) Click Add Filter and then close the filter dialog box.

(10) Continue on the next slide.

Slide 6


image


Clicking on the Bar Chart icon reveals that 6 of the 12 submissions took more than 10 minutes to complete. Chances are that these 6 submissions are those that recorded buildings with defects, since it takes more time to snap and annotate/mark photos of the defect on the building.

This is an instance that can be used for data quality assurance. For example, submissions of data collectors should be flagged as suspect if a form that ought to take about 5 minutes to complete is observed to take less than 1 minute to complete. This might suggest that the data collector is falsifying the data submissions and should prompt further investigation. Note that all data submitted by a suspect data collector can be analyzed or extracted by simply filtering the submissions using other metadata values e.g., Device ID, Subscriber ID etc.

On the other hand, a form observed to take longer than usual to complete might suggest that the form is badly designed and should be improved upon.

Slide 7


image


Analysis Question: How many of these submissions were uploaded in the month of February 2016?

(1) On the Dashboard click on Data Analysis -> Aggregates/Chart Based Analysis menu.

(2) Select the M and E Form in the Primary Form dropdown box.

(3) In the Group By Field dropdown box select the Survey Ack field. This field was chosen because all forms submitted have a similar value for this field.

(4) In the Form Fields section select the Count (Signboard) field or any other field.

(5) Click on the Filter icon as shown in the image above.

(6) In the dialog box presented, select Survey Date in the -Select form field- dropdown box.

(7) In the - Select Filter Operator - dropdown box select >.

(8) Click on the Filter Value textbox and select 31 January 2016.

(9) Click Add Filter.

(10) Make sure that Survey Date is still selected in the -Select filter field- dropdown box.

(11) In the Select Operator dropdown box select <.

(12) Click on the Filter Value textbox and select 1 March 2016.

(13) For the Filter Clause select the AND button.

(14) Click Add Filter and then close the filter dialog box.

(15) Continue on the next slide.

Slide 8


image


Another way to filter results based on submission date is to use the Submission Date value automatically recorded in the database when a form is submitted. This is a better way as its value does not depend on the device from which the submission is being made and therefore less prone to manipulation by data collectors. The image above shows a query filter example using the Submission Date.

Slide 9


image


Click on the Bar Chart icon to display a chart similar to what is shown above. From the chart, it is clear that 7 of the 12 submissions were made in February of 2016.

Slide 10


image


Analysis Request: I would like to see images of signboards captioned with the associated building codes and whether or not the building has a defect.

(1) On the Dashboard click on Data Analysis ->Photo Based Analysis menu.

(2) Select the M and E Form in the Primary Form dropdown box.

(3) In the -Select Annotation Field - dropdown box select the Facility Code and Any Defects fields.

(4) Select Signboard in the Form Field section.

(5) Click on the Get Photos icon to get the signboard photos.

Note that the photo display settings are configured such that image data is returned in groups of 30. To get the next 30 click on the Next Page button. To get the previous 30 images click on the Prev. Page button.

Slide 11


image


Analysis Request: I would like to see images of signboards captioned with the associated building codes and whether or not the building has a defect.

(1) On the Dashboard click on Data Analysis ->Photo Based Analysis menu.

(2) Select the M and E Form in the Primary Form dropdown box.

(3) In the -Select Annotation Field - dropdown box select the Facility Code and Any Defects fields.

(4) Select Signature Manager in the Form Field section.

(5) Click on the Get Photos icon to get the signboard photos.

As mentioned in the last slide, click on the Next Page button to get the next 30 images and Prev. Page button to get the last 30.

Slide 12


image


Analysis Request:I would like to see images of building defects captioned with the associated building code.

(1) On the Dashboard click on Data Analysis ->Photo Based Analysis menu.

(2) Select the M and E Form in the Primary Form dropdown box.

(3) In the -Select Annotation Field - dropdown box select the Facility Code and Any Defects fields.

(4) Select Defect Count [Defect Photo] in the Form Field section.

(5) Click on the Get Photos icon to get the signboard photos.

As mentioned in the last slide, click on the Next Page button to get the next 30 images and Prev. Page button to get the last 30.

This lesson is now complete.