What is Trend Analysis in Risk Management?
It is a common misnomer that trend analysis in risk management is used to forecast future performance. This understanding of trend analysis is both reductive and distorts the purpose of trend analysis in aviation SMS programs.
In aviation safety management systems, the one, thematic question that we need to ask over and over is:
- Is this an acceptable level of safety? (ALoS)
The real goals of trend analysis are to establish whether past performance is:
- Acceptable or unacceptable; and
- Moving in the right/wrong direction.
The assumption is that the trend will continue in the future (future forecast). The purpose is to know where in the safety program action needs to be taken in order to maintain ALoS.
Einstein supposedly said if he had an hour to solve a problem, he would spend the first 59 minutes understanding it. In the same way, a majority of trend analysis should be spent on preparation of data.
With trend analysis in risk management, the potential for establishing trends will only be as high as the quality of data at hand. Here are the basic elements needed for trend analysis.
1 – Gather Data for Trend Analysis
Gathering data may seem like a step that is so obvious it’s not worth mentioning. But data gathering techniques will make or break your ability to prepare data for trend analysis in risk management operations. Your trend analysis can only ever be as good as the data you gather and have available.
The ability to perform complex data analysis is one of the primary reasons organizations move to using aviation safety management software and/or an aviation safety database. Software and databases have vastly more potential than manual systems of data analysis. A database for data is just like a calculator for long division - it helps keep data
- Reliable; and
Manual data gathering simply takes significantly more time, and is at risk for errors. So the questions you need to ask yourself right now is:
- How are you getting data?
- Are you using metadata (more on this in the next section)?
If your data gathering is mostly automated (i.e., SMS software), then you already have a head start. If is manual, how are you ensuring the efficacy of your data gathering techniques? When you are manually entering reports, how much of the report is being input into a spreadsheet?
The purpose of gathering data for trend analysis is to have as much data and minutia available to analyze. During data analysis, data is your currency. The more, the better.
2 – Gather Metadata for Trend Analysis
Metadata is data about data. Why is metadata so important for trend analysis?
- It is a new layer of complexity for the data;
- It answers the who/what/when/where/why/how about data;
- Metadata should be the primary tool used for establishing trends; and
- In so many words, metadata gives your data context.
The best way to describe metadata is illustrate using an example. You have all your data, ready for analysis. Where do you start? Start breaking your data up with metadata:
- How many safety reports were reported at each hour of the day (when);
- How many safety reports were reported in each department (who);
- How many reported safety issues are there by type of issue (what); and
- How many issues were there per identified locations (where).
The list can go on. But breaking down data by the data by the data’s particular points (metadata), such as locations, reporters, type of report, time frame, etc., is a fantastic way to begin establishing trends. Mostly likely, you will start seeing trends right away based on the metadata.
Metadata is extremely hard to manage without the assistance of an aviation safety database or software. Possible, but difficult. Manual management of metadata requires heavy data entry during hazard reporting, such as manually entering in ALL minutia. Frankly, it’s much more cost effective for the SMS program (not to mention more reliable) to simply use a safety database or software for this.
3 – Establish Data Taxonomy
Data taxonomy is the grouping or classifying of data. In aviation SMS programs, data taxonomy translates to:
- Assessed risk level number/letter from risk matrix;
- Types of reports issues; and
- Most obviously, classifications of issues.
Data taxonomy is imported because it helps break your data into logical groups with which to establish trends. For example:
- In the past 12 months, how many high, medium, and low risk issues have been reported each month?
- How many of [classification type, i.e. bird strike] have been reported per month total, in the last 3 years?
- How many reported issue by type of report (i.e., safety, security, etc.) each month in last 12 months?
There are countless possibilities for using taxonomy to prepare data for trend analysis. Much like metadata, taxonomic data is a great jumping off point for trend analysis.
4 – Choose Time Frame to Establish Trends
As you work through trend analysis based on metadata, taxonomy, etc., it will be important that you use similar time frames in order to:
- Correlate different trends together;
- Understand overall scope of various trends; and
- Establish consistency in data analysis.
For example, simply be clear as to:
- Whether or not you are doing short or long term trend analysis; and
- What time frequency you will be using (days, weeks, months, quarter, etc.).
This data will ideally be presented in a visual chart showing your time frame on the X axis, and your data point(s) on the Y axis.
Final Though: Using Trend Data
Once you gather the above data, metadata, taxonomy, and timelines, you need to put your data together to begin trend analysis. Fortunately, the prep work is the most time consuming part. Once you have all of your data points, it's simply a matter of plugging in the data to your timeline.
As said, displaying trend data visually is definitely recommended. Using a line graph to display multiple potential trends is a useful way for comparing trends.
Similar to trend analysis, Shortfall Analysis is a new technique specifically designed to analysis SMS failures in safety incidents. Here is our free 16 page ebook covering what Shortfall Analysis is, and how to use it:
Image Credit: DarwinFish105