@thomasnield Yep, I saw JetBrains added a sample for Tensorflow in Kotlin/Native. Did you take a look at it BTW?

Neural networks framework that works in the browser, on JVM and natively is something I could invest my time on building. 😉

It looks like that is using a web service to process the image rather than using a local NN (although I didn’t check the code).

@chi The workflow in my role is heavily JDBC -> Tranformation -> Modeling (if any) ->TornadoFX. I use RxJava as the glue to tie everything together as data and event flows.

I'm still putting together this page which you might find helpful. https://github.com/thomasnield/kotlin-data-science-resources

Okay so apparently df.addColumn() is exactly what I was looking for. I spent an entire night finding that out when I could not find df.mutate() in Krangl.

@holgerbrandl I'm using krangl and wondering if there is a better way to alter a single column (keeping the rest as is). Currently I use

transmute

, but it seems I have to pass all unchanged columns as well to get them back.

Hi! I'd like to display some accelerometer data on an orthogonal mark (to show 4 curves of float values relative to a timestamp for each set of 4 values), along with a few lines to visualize the data. What I can do is put the accelerometer data in csv file or another persistable data structure. What would you recommend me to use to display this data for reasearch and device comparison purposes? I'd to do it with Kotlin of course

As an aside, does anyone have any machine learning books their recommend? What qualities make a ML book good? I know there have been significant improvements in how good we are at building neural networks in the last several years, so I have to imagine suggestions would need to be more recent?

Linear Programming with Kotlin Part III - Generating Multi-day Schedules http://tomstechnicalblog.blogspot.com/2018/02/linear-programming-with-kotlin-part-iii.html

Has anyone had success with Renjin? I got a project that would benefit from an R lib but would prefer to wrap Kotlin around it in a reasonably easy manner. Last time I tried to call R from Java, I thought about quitting programming and becoming a truck driver. http://www.renjin.org/

Hi all! It is a very early Spark support for Kotlin, but it works already. Working on Zeppelin support. https://github.com/khud/sparklin

Hi everyone, I'm reading the

Hands-On Machine Learning

book from O'Reilly and I'd like to rewrite some of those examples in Kotlin. Any suggestion on jvm libraries I could use as replacement for

Scikit-Learn

and its ecosystem?

tablesaw

seems quite interesting but maybe there's something more mature already available

Hey @ilya.chernikov & @apatrida, I’m presenting about Kotlin and Data Science next week at a Kotlin Night in Frankfurt, and I thought about including some slides about the kotlin kernel for jupyter. Unfortunately, it is in a rather broken state at the moment because of a dependency issue (see https://github.com/ligee/kotlin-jupyter/issues/15). Do you think there’s any chance that you could have a look, or guide me concerning a fix?

Missed the kotlin-night in Frankfurt? Here are my slides about Kotlin’s emerging data-science ecosystem https://holgerbrandl.github.io/kotlin4ds_kotlin_night_frankfurt/emerging_kotlin_ds_ecosystem.html

Is anyone using Kotlin with Spark? We're starting to and I ran into an issue with some helper functions not returning

Serializable

implementations. Thread starts here: https://kotlinlang.slack.com/archives/C1H43FDRB/p1521752154000105

Am I the only who found this user with several natural processing related repositories? https://github.com/KotlinNLP

I added a

NaiveBayesClassifier

to Kotlin-Statistics. New release coming shortly. https://github.com/thomasnield/kotlin-statistics#naive-bayes-classifier

OkAlgo 0.0.1 is now released. More extensions will come down the road. https://github.com/optimatika/okAlgo

Slides and resources from my talk at ChicagoJUG on Kotlin, Data Science, and Mathematical Modeling. https://github.com/thomasnield/kotlin_data_science_talk_ChicagoJUG/blob/master/README.md

@thomasnield What about deeplearning4j? Works pretty well with Kotlin for me.

@bjonnh what are you porting? The good news is ojAlgo is a pure JVM lib, so it should be Java-idiomatic already. I'd like to help on the documentation front and make it more fun to use with okAlgo. My hope one day is to copycat ojAlgo to Kotlin Native, as I've been pretty fond of its API so far with the optimizers.

a JS library to do quantum mechanics calculations

If you guys start an open source project, make sure to post a link here, for collaboration 🙂

Sparklin 0.2 is released! Fast new REPL engine, syntax highlight, customizable prompt and other features. Tested in production on cluster with CentOS Linux / Spark 1.6.3 / Hadoop 2.7. Experimental Kotlin support for Zeppelin is also added (without Spark right now). https://github.com/khud/sparklin

One of my bottlenecks is

fun kroneckerProduct_(x: MatrixStore<Double>, y: MatrixStore<Double>): MatrixStore<Double> {
    val newRows = x.countRows() * y.countRows()
    val newCols = x.countColumns() * y.countColumns()
    val store = SparseStore.PRIMITIVE.make(newRows, newCols)
    var count = 0

    for (i in 0 until x.countRows()) {
        for (k in 0 until y.countRows()) {
            for (j in 0 until x.countColumns()) {
                for (l in 0 until y.countColumns()) {
                    store.set((count/newRows)%newCols, count%newRows , x.get(i, j) * y.get(k, l))
                    count++
                }
            }
        }
    }
    return store
}