insights.jpg

Elite Data Analysis For Security III: Emphasizing Data Quality

By Dan Jodarski

How To Examine Your Data Quality When Conducting Data Analysis For SecurityIn order to keep your federal agency secure, data science is now a must. 

But not all data is created equal. 

In order to make the most well informed decisions for agency security, your data needs to be accurate, up to date and properly portrayed. A pretty dashboard or fancy graphic display is no substitute for a solid security decision – which means data quality should be one of your highest priorities. 

In this third post of a five-part series, we drill into why your agency security operation should care about data quality, and how to determine the accuracy of your data before you make a critical decision based on it. 

In Part I, we discussed robust data collection tactics for stronger security decision-making, and in Part II, we examined how to avoid common pitfalls when working with data dashboards

Understanding the quality of your data requires you to dig beneath the surface layer of your data dashboard and investigate the underlying story behind your data. Without assessing data quality, your agency security decisions are made in the dark. 

Here are three steps your defense or intelligence agency needs to take to increase your data quality and bolster your confidence in data-driven decisions: 

1. Eradicate Human Errors by Automating Where Possible

Conclusions drawn from data analysis are always prone to some level of human error. However, you should work to eliminate as many error-causing behaviors as possible, especially regarding data quality. As a general rule, machines tend to be faster and more accurate performing repeatable tasks following a clear logic. Simple changes like writing a program to enter data in the backend of your dashboard, rather than rely on typo-prone administrators to enter it manually can reduce confounding data errors. 

When it comes to data collection, identify which manual processes produce data errors and automate where possible. 

2. Strive for Reproducibility  

You may not be able to automate all your sound data collection tactics, but you definitely need to reproduce them. Make sure any data retrieving, data cleaning, and data analysis tasks conducted in creating your dashboard are captured in writing, using tools like markdown and GitHub that encourage best practices for reproducible data science.

Also, remember data science may be an essential function of your job as an agency security or suitability executive, but you don’t have to be a full-blown data scientist. Getting a professional advisor or contractor to create your data dashboards is a wise decision, as long as you don’t let an outside partner hide their methods from you.  

3. Identify Outliers And The Story Behind Them 

Your dashboard should allow you to easily identify outlying data points and examine the story behind them. Outliers could be a sign that your new security strategy or operational practice is producing exceptional results. Or, outliers might signal a data quality issue in how a variable is being measured. 

If you aren’t getting a consistent measurement or if your margin of error is too wide, then you won’t be able to make data-driven decisions from your information. Using data analysis methods like monitoring standard deviation and identifying spurious correlations, will help you understand the true story your data is trying to tell you. 

.High-quality decisions are based on high-quality data. Increase your data quality today and your agency security decisions will become more reliable tomorrow.  

Does your federal agency need to improve organizational security operations within a limited schedule and budget? Click below to download this e-book from Big Sky Associates and discover how to make process improvement efforts that are cost-effective for your budget. 

Download Your Free Report: The Ultimate Process Improvement Guide From Initial Data Analysis To Final Implementation Plan

Catch up with the rest of the elite data analysis series: