And very simple statistics can be done in FileMaker without the need for an additional tool so much depends on what you want to accomplish as to what tool, if any you need to reach your goal.
What is the best data analytics tool
"best" is always subjective. "effective" depends on both the design of the solution and the nature of the analytics you are after.
Can you give some examples of reports that you are after?
If you mean real Data Analytics, I would suggest exporting the data to (or connecting with, from FMP) R, SPSS or tools built for what you might need.
I also use simple REST micro-services where I code stat routines using powerful APIs like Apache Commons which go so well beyond FMP (computation of p-values, hypothesis testing, regression, ANOVA, etc.). These APIs are FAST, RELIABLE, and are in widespread use. These APIs are extremely easy to use as well.
Using the micro-serivce approach and FMP's INSERT FROM URL (or curl), it's also straightforward to connect to R (dynamically) and use R's immense data analytic's capabilities. But in this context, I connect to the REST service real-time directly from FMP and THEN from the REST service connect to the other programs for which there is a well-defined (FREE) API for Java.
FMP Example:, I could have two fields on a FMP layout with the number 1 and in the second field the number 6. The meaning would be clear on the layout, but the "1" is the number of occurrences expected with the mean of 6. Then with a quick call to R via the micro-service (again, using FMP's INSERT FROM URL), I can then (from Java in the micro-service) call R (running as a service on the machine or server) with this command: dpois(1,6) for a Poisson test. (R returns: 0.01487251 to FMP in this case meaning there is roughly a 14% chance of exactly one occurrence).
So, for serious data analytics, you could still use FMP, but I only use it as a front end and then only if there is an existing FMP application which needs some analytics "bolted on".
No need for expensive add-ons. R is free. Apache Commons is free. Micro-services are free.
It's nice that FMP gives you two ways that I know of to leverage these external mechanisms (INSERT FROM URL (script) and CURL (CF)).
I could give you more detailed examples if that would be helpful.
1 of 1 people found this helpful
Since this thread was not yet marked as "answered", I went ahead and did a quick FMP application that connects to the external statistics library, R to do a simple Poisson statistic. (Note: In a real application, I would first check to make sure the data were from a Poisson distribution using a "Goodness of Fit" test.).
For this code to work, you would need to have R installed on the computer and be running RServe, which could be started, if necessary, dynamically, at run-time (outside of FMP).
You can download the free R statistical environment here: R: The R Project for Statistical Computing
KEY BENEFITS OF THE SERVICE APPROACH:
1. You can use the service from any application that can do HTTP GETs (POST, ....)
2. Thus, not limited to FMP, but FMP can connect to it -- see example 1 below
3. No (expensive) FMP third-party plug-ins needed (if they exist)
4. No FMP plug-in configuration chores across multiple machines/servers (keep installed, up to date, etc.) for third-party plug-ins
6. Not difficult to create
FileMaker Example -- Cool! (results come back nearly instantly)
(Below, I used FMP's INSERT FROM URL to connect to the microservice using the same URL as shown in other examples below.)
In the browser:
Call like this: "http://localhost:4567/poisson/1/6 "
Result in browser window: 0.014872513059998146
In a "DOS Window" (Windows) or Terminal (Mac) using CURL:
Not sure this information is useful for you, but please mark the discussion answered if you're set.
HOPE THIS HELPS.