The Splunk Machine Learning Toolkit is packed with machine learning algorithms, new visualizations, web assistant and much more. This blog sheds light on some features and commands in Splunk Machine Learning Toolkit (MLTK) or Core Splunk Enterprise that are lesser known and will assist you in various steps of your model creation or development. With each new release of the Splunk or Splunk MLTK a catalog of new commands are available. I attempt to highlight commands that have helped in some data science or analytical use-cases in this blog.Read more
There are multiple (almost discretely infinite) methods of outlier detection. In this blog I will highlight a few common and simple methods that do not require Splunk MLTK (Machine Learning Toolkit) and discuss visuals (that require the MLTK) that will complement presentation of outliers in any scenario. This blog will cover the widely accepted method of using averages and standard deviation for outlier detection. The visual aspect of detecting outliers using averages and standard deviation as a basis will be elevated by comparing the timeline visual against the custom Outliers Chart and a custom Splunk’s Punchcard Visual.
Some Key Concepts
Understanding some key concepts are essentials to any Outlier Detection framework. Before we jump into Splunk SPL (Search Processing Language) there are basic ‘Need-to-know’ Math terminologies and definitions we need to highlight:
- Outlier Detection Definition: Outlier detection is a method of finding events or data that are different from the norm.
- Average: Central value in set of data.
- Standard Deviation: Measure of spread of data. The higher the Standard Deviation the larger the difference between data points. We will use the concept of standard substantially in today’s blog. To view the manual method of standard deviation calculation click here.
- Time Series: Data ingested in regular intervals of time. Data ingested in Splunk with a timestamp and by using the correct ‘props.conf’ can be considered “Time Series” data
Additionally, we will leverage aggregate and statistic Splunk commands in this blog. The 4 important commands to remember are:
- Bin: The ‘bin’ command puts numeric values (including time) into buckets. Subsequently the ‘timechart’ and ‘chart’ function use the bin command under the hood
- Eventstats: Generates statistics (such as avg,max etc) and adds them in a new field. It is great for generating statistics on ‘ALL’ events
- Streamstats: Similar to ‘stats’ , streamstats calculates statistics at the time the event is seen (as the name implies). This feature is undoubtedly useful to calculate ‘Moving Average’ in additional to ordering events
- Stats: Calculates Aggregate Statistics such as count, distinct count, sum, avg over all the data points in a particular field(s)
The data used in this blog is Splunk’s open sourced “Bots 2.0” dataset from 2017. To gain access to this data please click here. Downloading this data set is not important, any sample time series data that we would like to measure for outliers is valid for the purposes of this blog. For instance, we could measure outliers in megabytes going out of a network OR # of logins in a applications using the using the same type of Splunk query. The logic used to the determine outliers is highly reusable.
There are four methods commonly seen methods applied in the industry for basic outlier detection. They are in the sections below:
1. Using Static Values
The first commonly used method of determining an outlier is by constructing a flat threshold line. This is achieved by creating a static value and then using logic to determine if the value is above or below the threshold. The Splunk query to create this threshold is below :
<your spl base search> … | timechart span=6h sum(mb_out) as mb_out | eval threshold=100 | eval isOutlier=if('mb_out' > threshold, 1, 0)
2. Average with Static Multiplier
In addition to using arbitrary static value another method commonly used method of determining outliers, is a multiplier of the average. We calculate this by first calculating the average of your data, following by selecting a multiplier. This creates an upper boundary for your data. The Splunk query to create this threshold is below:
<your spl base search> … | timechart span=12h sum(mb_out) as mb_out | eventstats avg("mb_out") as average | eval threshold=average*2 | eval isOutlier=if('mb_out' > threshold, 1, 0)
3. Average with Standard Deviation
Similar to the previous methods, now we use a multiplier of standard deviation to calculate outliers. This will result in a fixed upper and lower boundary for the duration of the timespan selected. The Splunk query to create this threshold is below:
<your spl base search> ... | timechart span=12h sum(mb_out) as mb_out | eventstats avg("mb_out") as avg stdev("mb_out") as stdev | eval lowerBound=(avg-stdev*exact(2)), upperBound=(avg+stdev*exact(2)) | eval isOutlier=if('mb_out' < lowerBound OR 'mb_out' > upperBound, 1, 0)
Notice that with the addition of the lower and upper boundary lines the timeline chart becomes cluttered.
4. Moving Averages with Standard Deviation
In contrast to the previous methods, the 4th most common method seen is by calculating moving average. In short, we calculate the average of data points in groups and move in increments to calculate an average for the next group. Therefore, the resulting boundaries will be dynamic. The Splunk search to calculate this is below:
<your spl base search> ... | timechart span=12h sum(mb_out) as mb_out | streamstats window=5 current=true avg("mb_out") as avg stdev("mb_out") as stdev | eval lowerBound=(avg-stdevexact(2)), upperBound=(avg+stdevexact(2)) | eval isOutlier=if('mb_out' < lowerBound OR 'mb_out' > upperBound, 1, 0)
Tips: Notice the “isOutliers” line in the timeline chart, in order to make smaller values more visible format the visual by changing the scale from linear to log format.
Using the MLTK Outlier Visualization
Splunk’s Machine Learning Toolkit (MLTK) contains many custom visualization that we can use to represent data in a meaningful way. Information on all MLTK visuals detailed in Splunk Docs. We will look specifically at the ‘Outliers Chart’. At the minimum the outlier chart requires 3 additional fields on top of your ‘_time’ & ‘field_value’. First, would need to create a binary field ‘isOutlier’ which carries the value of 1 or 0, indicating if the data point is an outlier or not. The second and third field are ‘lowerBound’ & ‘upperBound’ indicating the upper and lower thresholds of your data. Because the outliers chart trims down your data by displaying only the value of data point and your thresholds, we can conclude through use that it is clearer and easier to understand manner. As a recommendation it should be incorporated in your outliers detection analytics and visuals when available.
Continuing from the previous paragraph, take a look at the below snippets at how the impact the outliers chart is in comparison to the timeline chart. We re-created the same SPL but instead of applying timeline visual applied the ‘Outliers Chart’ in the same order:
|Cleaner presentation and less clutter
|You need to install Splunk MLTK (and its pre-requisites) to take advantage of the outliers chart
|Easier to understand as determining the boundaries becomes intuitive vs figuring out which line is the upper or lower threshold
|Unable to append additional fields in the Outliers chart
Adding Depth to your Outlier Detection
Determining the best technique of outlier detection can become a cumbersome task. Hence, having the right tools and knowledge will free up time for a Splunk Engineer to focus on other activities. Creating static thresholds over time for the past 24hrs, 7 days, 30 days may not be the best approach to finding outliers. A different way to measure outliers could be by looking at the trend on every Monday for the past month or 12 noon everyday for the past 30 days. We accomplish this by using two simple and useful eval functions:
| eval HourOfDay=strftime(_time, "%H") | eval DayOfWeek=strftime(_time, "%A")
Using Eval Functions in SPL
Continuing from the previous section, we incorporate the two highlighted eval functions in our SPL to calculate the average ‘mb_out’. However, this time the average is based on the day of the week and the hour of the day. There are a handful of advantages of this method:
- Extra depth of analysis by adding 2 additional fields you can split the data by
- Intuitive method of understanding trends
Some use cases of using the eval functions are as follows:
- Network activity analysis
- User behaviour analysis
Visualizing the Data!
We will focus on two visualizations to complement our analysis when utilizing the eval functions. The first visual, discussed before, is the ‘Outliers Chart’ which is a custom visualization in Splunk MLTK. The second visual is another custom visualization ‘PunchCard’, it can be downloaded from Splunkbase here (https://splunkbase.splunk.com/app/3129/).
The outliers chart has a feature which results in a ‘swim lane’ view of a selected field/dimension and your data points while highlighting points that are outliers. To take advantage of this feature, we will use a Macro “splitby” which creates a hidden field(s) “_<Field(s) you want data to split by>”. The rest of the SPL is shown below
< your base SPL search >
... | eventstats avg("mb_out") as avg stdev("mb_out") as stdev by "HourOfDay" | eval avg=round(avg,2) | eval stdev=round(stdev,2) | eval lowerBound=(avg-stdev*exact(2)), upperBound=(avg+stdev*exact(2)) | eval isOutlier=if('mb_out' < lowerBound OR 'mb_out' > upperBound, 1, 0) | `splitby("HourOfDay")` | fields _time, "mb_out", lowerBound, upperBound, isOutlier, * | fields - _raw source kb* byt* | table _time "mb_out" lowerBound upperBound isOutlier *
This search results in an Outlier Chart that looks like this:
The Outliers Chart has the capability to split by multiple fields, however in our example splitting it by a single dimension “HourOfDay” is sufficient to show its usefulness.
The PunchCard visual is the second feature we will use to visualize outliers. It displays cyclical trends in our data by representing aggregated values of your data points over two dimensions or fields. In our example, I’ve calculated the sum of outliers over a month based on “DayOfWeek” as my first dimension and “HourOfDay” as my second dimension. I’ve adding the outliers of these two fields and displaying it using the PunchCart visual. The SPL and image for this visual is show below:
< your base SPL search > ... | streamstats window=10 current=true avg("mb_out") as avg stdev("mb_out") as stdev by "DayOfWeek" "HourOfDay"
| eval avg=round(avg,2)
| eval stdev=round(stdev,4)
| eval lowerBound=(avg-stdevexact(2)), upperBound=(avg+stdevexact(2))
| eval isOutlier=if('mb_out' < lowerBound OR 'mb_out' > upperBound, 1, 0)
| stats sum(isOutlier) as mb_out by DayOfWeek HourOfDay
| table HourOfDay DayOfWeek mb_out
Summary and Wrap Up
Trying to find outliers using Machine Learning techniques can be a daunting task. However I hope that this blog gives an introduction on how you can accomplish that without using advanced algorithms. Consequently, using basic SPL and built-in statistic functions can result in visuals and analysis that is easier for stakeholders to understand and for the analyst to explain. So summarizing what we have learnt so far:
- One solution does not fit all. There are multiple methods of visualizing your analysis and exploring your result through different visual features should be encouraged
- Use Eval functions to calculate “DayOfWeek” and “HourOfDay” wherever and whenever possible. Adding these two functions provides a simple yet powerful tool for the analyst to explore the data with additional depth
- Trim or minimize the noise in your Outliers visual by using the Outliers Chart. The chart is beneficial in displaying only your boundaries and outliers in your data while shaving all other unnecessary lines
- Use “log” scale over “linear” scale when displaying data with extremely large ranges
© Discovered Intelligence Inc., 2020. Unauthorised use and/or duplication of this material without express and written permission from this site’s owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Discovered Intelligence, with appropriate and specific direction (i.e. a linked URL) to this original content.
Part II of the Forecasting Time Series blog provides a step by step guide for fitting an ARIMA model using Splunk’s Machine Learning Toolkit. ARIMA models can be used in a variety of business use cases. Here are a few examples of where we can use them:
- Detecting anomalies and their impact on the data
- Predicting seasonal patterns in sales/revenue
- Streamline short-term forecasting by determine confidence intervals
From Part 1 of the blog series, we identified how you can use Kalman Filter for forecasting. The observation we made from the resulting graphs demonstrated how it was also useful in reducing/filtering noise (which is how it gets its name ‘Filter’) . On the other hand ARIMA belongs to a different class of models. In comparison to a Kalman filter, ARIMA models works on data that has moving averages over time or where the value of a data point is linearly depending on its previous value(s). In these two scenarios it makes more sense to use ARIMA over Kalman Filter. However good judgement, understanding of the data-set and objective of forecasting should always be the primary method of determining the algorithm.
Part II of this blog series aims to familiarize a Splunk user using the MLTK Assistant for forecasting their time series data, particularly with the ARIMA option. This blog is intended as a guide in determining the parameters and steps to utilize ARIMA for your data. In fact, it is a generalized template that can be used with any processed data to forecasting with ARIMA in Splunk’s MLTK. An advantage of using Splunk for forecasting is its benefit in observing the raw data side by side with the predicted data and once the analysis is complete, a user can create alerts or other actions based on a future prediction. We will talk more about creating alerts based on predicted or forecasted data in a future blog (see what I predicted there ;)?)
Fundamental Concept for ARIMA Forecasting
A fundamental concept to understand before we move ahead with ARIMA is that the model works best with stationary data. Stationary data has a constant trend that does not change overtime. The average value is also independent of time as another characteristic of stationary data.
A simple example of non-stationary data is are the two graphs below, the first without a trendline, the second with a yellow trendline to show an average increase in the value of our data points. The data needs to be transformed into stationary data to remove the increasing trend.
Using Splunk’s autoregress command we can apply differencing to our data. The results are immediately visible through line chart visual! The below command can be used on any time series data set to demonstrate differencing.
… | autoregress value | eval new_value=value-value_p1 | fields _time new_value
Without creating a trendline for the below graph we can see that the data fluctuates around a constant mean value of ‘0’, we can say that differencing is applied. Differencing to make the data stationary can increase the accuracy and fit of our ARIMA forecast. To read more about differencing and other rules that apply on ARIMA, navigate to the Duke URL provided in the useful link section:
Differencing is simply subtracting the current and previous data points. In our example we are only applying differencing by an order of 1, meaning we will subtract the present data point by one data point in reverse chronological order. There are different types of non-stationary graphs, which require in-depth domain knowledge of ARIMA, however we simplify it in this blog and use differencing to remove the non-constant trend in this example 😊!
From part 1 of this blog series we can see that our data does not have a constant trend, as a result we apply differencing to our dataset. The step to apply differencing from the MLTK Assistant is detailed in the ‘Determining Starting Points’ section. Differencing in ARIMA allows the user to see spikes or drops (outliers) in a different perspective in comparison to Kalman Filter.
Walkthrough of MLTK Assistant for ARIMA
ARIMA is a popular and robust method of forecasting long-term data. From blog 1 we can describe Kalman Filter’s forecasting capabilities as extending the existing pattern/spikes, sort of a copy-paste method which may be advantageous when forecasting short-term data. ARIMA has an advantage in predicting data points when the we are uncertain about the future trend of the data points in the long-term. Now that we have got you excited about ARIMA, lets see how we can use it in Splunk’s MLTK!
We use the Machine Learning Toolkit Assistant for forecasting timeseries data in Splunk. Navigate to the Forecast Time Series Assistant page (Under the Classic Menu option) and use the Splunk ‘inputlookup’ command to view the process_time.csv file.
Once we add the dataset click on Algorithm and select ‘ARIMA’ (Autoregressive Integrated Moving Average), and ‘value’ as your field to forecast. You will notice that the ARIMA arguments will appear.
There are three arguments that make up the ARIMA model:
|AutoRegressive – p
|Auto regressive (AR) component refers to the use of past values in the regression equation. Higher the value the more past terms you will use in the equation. This concept is also called ‘lags’. Another way of describing this concept is if the value your data point is depending on its previous value e.g process time right now will depend on the process time 30 seconds before (from our data set)
|Integrated – d
|The d represents the degrees of differencing as discussed in the previous section. This makes up the integrated component of the ARIMA model and is needed for the stationary assumption of the data.
|Moving Average – q
|Moving Average in ARIMA refers to the use of past errors in the equation. It is the use of lagging (like AR) but for the error terms.
Determine Starting Points
Identify the Order of Differencing (d)
As a refresher, we utilized the same dataset we worked with in part 1 of the blog series regarding the Kalman filter. As I input my process_time.csv file in the assistant, I enter the future_timespan variable as 20 and the holdback as 20. I’ve kept the confidence interval as default value ‘95’. Once the argument values are populated click on ‘Forecast’ to see the resulting graphs.
As a note, my ARIMA arguments described above are ARIMA(0,0,0) which can represented as a mathematics function ARIMA(p,d,q), where p,d,q = 0. We use this functional representation of the variables frequently in this blog for consistently with generally used mathematical languages.
When we click on forecast, observe the line chart graph from the results that show. This above graph confirms that the data is non-stationary, we will apply differencing to make it stationary. We can accomplish this by increasing the value of our ‘d’ argument from ‘0’ to ‘1’ in the forecasting assistant and clicking on forecast again. This step is essential to meet one of the main criteria’s of using ARIMA discussed in the ‘Fundamental Concept for ARIMA’ section.
Identifying AR(p) and MA(q)
After we apply differencing to our data our next step is to determine the AR or MR terms that mitigate any auto correlation in our data. There are two popular methods of estimating the these two parameters. We will expand on one of the methods in this blog.
The first method for estimating the value of ‘p’ and ‘q’ is to use the Akaiki Information Criteria (AIC) and the Baysian Information Criteria (BIC), however using them is outside the scope of the blog as we will use a different method from the MLTK given the tools we have at hand. For the curious mind, the following blog contains detailed information on AIC and BIC to determine our ‘p’ and ‘q’ values:
After we have applied differencing to our time series data, we review the PCAF and the ACF plots to determine an order for AR(q) or MA(q). We will apply ARIMA(0,1,0) in our ARIMA MLTK assistant and then click on ‘Forecast’ to view the results of the graph. The below image shows the values that we entered in the assistant:
Once we click on forecast, we view the PACF plot to estimate a value for AR(p) model. Similarly we use the ACF plot to estimate a value for MA(q). The graphs are shown in the screenshot below.
We examine the PACF plot for a suggestion for our AR value, by counting the prominent high spikes. From the plot below I’ve circled the prominent spikes in the PACF graph. The value of AR (p) that we pick is 4.
We examine the ACF plot for a suggestion for our MA value, by counting the prominent high spikes. From the plot below I’ve circled the prominent spikes in the ACF graph. The value of AR (q) that we pick is 5.
We can now add in the values for the parameter integrated (d) – 1 and our estimates for AR – 4, and MA -5 in the Splunk MLTK. Once added in the assistant, click on ‘Forecast’.
For this particular combination for values we can see that once we click on ‘Forecast’, we get an error regarding the ‘invertability’ of the dataset as shown in the screenshot below. Without going too deep into the mathematics, it means that our model does not converge when it forecasts. I’ve added a link in the references and links section at the end for your interest! This error can be resolved by adjusting the values of model, similar to a ‘trail an error’ approach explained in the next section.
Optimize Your P and Q Values
Estimating this method of AR and MA is subjective to what can be considered as ‘prominent spikes’, this can result in estimating values of ‘q’ and ‘p’ that are not an optimal fit for the data. To resolve this we constructed a table displaying the R-squared and Root Mean Square Error (RMSE) values from the model error statistics from the MLTK assistance, for each combination of ‘p’ and ‘q’. An empty cell indicates an invertability error, while the other cells contain the value of R-squared and RMSE.
A higher R-squared indicates a better fit the model has on the data. R-squared is the amount of variability that the model can explain on the process time data points.
On the other hand, the lower the RMSE is the better the fit of the model. Root mean square is the difference between the data points the model predicted and our holdback points from the raw data.
We pick values of ‘p’ and ‘q’ that minimize RMSE and maximize R-square as the best fit to our data. From the table below we can see that q=5 and p=5 optimize the prediction for us.
|Integrated (d) = 0
|Moving Average (q)
|R2 Stat: -0.0015 RMSE: 19.31
|R2 Stat: 0.1976 RMSE: 16.35
|R2 Stat: 0.1977 RMSE: 16.34
|R2 Stat: 0.2699 RMSE: 15.60
|R2 Stat: 0.2696 RMSE: 15.60
|R2 Stat: 0.3114 RMSE: 15.14
|R2 Stat: 0.2401 RMSE: 15.91
|R2 Stat: 0.2486 RMSE: 15.82
|R2 Stat: 0.2780 RMSE: 15.51
|R2 Stat: 0.2329 RMSE: 15.98
|R2 Stat: 0.4053 RMSE: 14.07
|R2 Stat: 0.2452 RMSE: 15.85
|R2 Stat: 0.3017 RMSE: 15.25
|R2 Stat: 0.3214 RMSE: 15.03
|R2 Stat: 0.2872 RMSE: 15.41
|R2 Stat: 0.4185 RMSE: 13.92
|R2 Stat: 0.4428 RMSE: 13.62
|R2 Stat: RMSE:
|R2 Stat: 0.4343 RMSE: 13.72
|R2 Stat: 0.4456 RMSE: 13.58
|R2 Stat: 02826 RMSE: 15.46
|R2 Stat: 0.4185 RMSE: 13.92
|R2 Stat:0.3241 RMSE: 15.00
|R2 Stat: 0.2826 RMSE: 15.46
|R2 Stat: 0.3133 RMSE: 15.99
|R2 Stat: 0.4385 RMSE: 13.67
|R2 Stat: 0.4515 RMSE: 13.52
Viewing Your Results
Once we have picked the values of p and q that optimize our model, we can go ahead plug the numbers in our assistant and click on forecast to display the forecasted graph. The values to plug in the assistant are as follows: p-5, d-1, q-5, holdback-20, forecast-20. The screenshots below show the values entered in the assistant and the resulting forecast graph.
A this point many would be satisfied with the forecast as the visual of the data itself is enough to analyse, asses and then make a judgement on the action(s) to take. The next step details how you can view the data and lists some ideas of alerts that can be constructed
We can view the SPL used powering the graph by either clicking on ‘Open in Search’ or ‘ ‘Show SPL’. I prefer the ‘Open in Search’ option as it automatically open a new tab, allowing me to further understand how the SPL is constructed in the forecast and to view the data. Once a tab browser tab opens click on the ‘statistics’ option to view the raw data points, predicted data points and the confidence intervals created by our model. I have added the SPL from the image for your convenience below:
| inputlookup process_time.csv | fit ARIMA _time value holdback=20 conf_interval=95 order=5-1-5 forecast_k=40 as prediction | `forecastviz(40, 20, "value", 95)`
I added another filter to my SPL to only view the forecasted process data from the ARIMA model as shown below:
| inputlookup process_time.csv | fit ARIMA _time value holdback=20 conf_interval=95 order=5-1-5 forecast_k=40 as prediction | `forecastviz(40, 20, "value", 95)` | search "lower95(prediction)"=*
The resulting table lists all the necessary data in a clean tabular format (that we are all familiar with) for creating alerts based on our predicted process time. Here are some ideas on creating alerts based on the data we worked with:
- Create alert when the predicted value of the process time goes above a certain threshold
- Create alert when the average process time over a timespan is predict to stay above normal limits
- Create alert based on outlier detection, when the predicted data is outside the lower or upper boundaries
Creating alerts based on our predict data allows us to be proactive of potential increase or decrease of our input variable
Summarizing ARIMA Forecasting in MLTK
Lets summarize what we have discussed so far in this blog:
- A mathematical prerequisites of the model
- Determining differencing requirement
- Determine starting values for AR() and MA()
- Optimize your AR() and MA() values based on error statistics
- Forecast your data based on values decided in Step 4
- View data and determine any alerts conditions
Prior to the above steps, we need to ensure that our data has been pre-processed or transformed in a MLTK-friendly manner. The pre-process steps include but not limited to; ensuring no gaps in the time series data, determine the relevance of data to forecasting, group data in time intervals (30 second, 1 minute etc). The pre-processing steps are important to create uniformity in the data input allow Splunk’s MLTK to analyse and forecast your data.
Hopefully this blog, streamlines the process of forecasting using ARIMA in Splunk’s MLTK. There are limitations as with any algorithm on forecasting using this method, as it involves a more theoretical knowledge in mathematics I’ve added two links in the the useful links section (first link is navigates you to on ‘datascienceplus.com’ and the second to ’emeraldinsight.com’) to further read on them.
© Discovered Intelligence Inc., 2019. Unauthorised use and/or duplication of this material without express and written permission from this site’s owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Discovered Intelligence, with appropriate and specific direction (i.e. a linked URL) to this original content.
In this blog we will use a classification approach for predicting Spam messages. A classification approach categorizes your observations/events in discrete groups which explain the relationship between explanatory and dependent variables which are your field(s) to predict. Some examples of where you can apply classification in business projects are: categorizing claims to identify fraudulent behaviour, predicting best retail location for new stores, pattern recognition and predicting spam messages via email or text. Read more
The Splunk Machine Learning Toolkit (MLTK) assists in applying machine learning techniques and methods against your data. This article discusses how to get started with the MLTK including installation and some initial testing and examples. Read more