RStudio AI Weblog: TensorFlow function columns: Remodeling your information recipes-style


It’s 2019; nobody doubts the effectiveness of deep studying in pc imaginative and prescient. Or pure language processing. With “regular,” Excel-style, a.ok.a. tabular information nonetheless, the state of affairs is totally different.

Mainly there are two circumstances: One, you’ve numeric information solely. Then, creating the community is simple, and all shall be about optimization and hyperparameter search. Two, you’ve a mixture of numeric and categorical information, the place categorical may very well be something from ordered-numeric to symbolic (e.g., textual content). On this latter case, with categorical information coming into the image, there may be an especially good concept you may make use of: embed what are equidistant symbols right into a high-dimensional, numeric illustration. In that new illustration, we are able to outline a distance metric that permits us to make statements like “biking is nearer to working than to baseball,” or “😃 is nearer to 😂 than to 😠.” When not coping with language information, this system is known as entity embeddings.

Good as this sounds, why don’t we see entity embeddings used on a regular basis? Effectively, making a Keras community that processes a mixture of numeric and categorical information used to require a little bit of an effort. With TensorFlow’s new function columns, usable from R by a mixture of tfdatasets and keras, there’s a a lot simpler approach to obtain this. What’s extra, tfdatasets follows the favored recipes idiom to initialize, refine, and apply a function specification %>%-style. And at last, there are ready-made steps for bucketizing a numeric column, or hashing it, or creating crossed columns to seize interactions.

This publish introduces function specs ranging from a situation the place they don’t exist: principally, the established order till very just lately. Think about you’ve a dataset like that from the Porto Seguro automotive insurance coverage competitors the place a few of the columns are numeric, and a few are categorical. You wish to practice a completely related community on it, with all categorical columns fed into embedding layers. How will you try this? We then distinction this with the function spec manner, which makes issues so much simpler – particularly when there’s a variety of categorical columns. In a second utilized instance, we show the usage of crossed columns on the rugged dataset from Richard McElreath’s rethinking package deal. Right here, we additionally direct consideration to a couple technical particulars which might be price figuring out about.

Mixing numeric information and embeddings, the pre-feature-spec manner

Our first instance dataset is taken from Kaggle. Two years in the past, Brazilian automotive insurance coverage firm Porto Seguro requested individuals to foretell how seemingly it’s a automotive proprietor will file a declare based mostly on a mixture of traits collected throughout the earlier yr. The dataset is relatively massive – there are ~ 600,000 rows within the coaching set, with 57 predictors. Amongst others, options are named in order to point the kind of the info – binary, categorical, or steady/ordinal. Whereas it’s widespread in competitions to attempt to reverse-engineer column meanings, right here we simply make use of the kind of the info, and see how far that will get us.

Concretely, this implies we wish to

  • use binary options simply the way in which they’re, as zeroes and ones,
  • scale the remaining numeric options to imply 0 and variance 1, and
  • embed the specific variables (each by itself).

We’ll then outline a dense community to foretell goal, the binary end result. So first, let’s see how we might get our information into form, in addition to construct up the community, in a “handbook,” pre-feature-columns manner.

When loading libraries, we already use the variations we’ll want very quickly: Tensorflow 2 (>= beta 1), and the event (= Github) variations of tfdatasets and keras:

On this first model of making ready the info, we make our lives simpler by assigning totally different R varieties, based mostly on what the options characterize (categorical, binary, or numeric qualities):

# downloaded from https://www.kaggle.com/c/porto-seguro-safe-driver-prediction/information
path <- "practice.csv"

porto <- read_csv(path) %>%
  choose(-id) %>%
  # to acquire variety of distinctive ranges, later
  mutate_at(vars(ends_with("cat")), issue) %>%
  # to simply hold them other than the non-binary numeric information
  mutate_at(vars(ends_with("bin")), as.integer)

porto %>% glimpse()
Observations: 595,212
Variables: 58
$ goal         <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0,…
$ ps_ind_01      <dbl> 2, 1, 5, 0, 0, 5, 2, 5, 5, 1, 5, 2, 2, 1, 5, 5,…
$ ps_ind_02_cat  <fct> 2, 1, 4, 1, 2, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1,…
$ ps_ind_03      <dbl> 5, 7, 9, 2, 0, 4, 3, 4, 3, 2, 2, 3, 1, 3, 11, 3…
$ ps_ind_04_cat  <fct> 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1,…
$ ps_ind_05_cat  <fct> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_06_bin  <int> 0, 0, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_07_bin  <int> 1, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 1,…
$ ps_ind_08_bin  <int> 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0,…
$ ps_ind_09_bin  <int> 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0,…
$ ps_ind_10_bin  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_11_bin  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_12_bin  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_13_bin  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_14      <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_15      <dbl> 11, 3, 12, 8, 9, 6, 8, 13, 6, 4, 3, 9, 10, 12, …
$ ps_ind_16_bin  <int> 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0, 0,…
$ ps_ind_17_bin  <int> 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_ind_18_bin  <int> 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 1,…
$ ps_reg_01      <dbl> 0.7, 0.8, 0.0, 0.9, 0.7, 0.9, 0.6, 0.7, 0.9, 0.…
$ ps_reg_02      <dbl> 0.2, 0.4, 0.0, 0.2, 0.6, 1.8, 0.1, 0.4, 0.7, 1.…
$ ps_reg_03      <dbl> 0.7180703, 0.7660777, -1.0000000, 0.5809475, 0.…
$ ps_car_01_cat  <fct> 10, 11, 7, 7, 11, 10, 6, 11, 10, 11, 11, 11, 6,…
$ ps_car_02_cat  <fct> 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1,…
$ ps_car_03_cat  <fct> -1, -1, -1, 0, -1, -1, -1, 0, -1, 0, -1, -1, -1…
$ ps_car_04_cat  <fct> 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 8, 0, 0, 0, 0, 9,…
$ ps_car_05_cat  <fct> 1, -1, -1, 1, -1, 0, 1, 0, 1, 0, -1, -1, -1, 1,…
$ ps_car_06_cat  <fct> 4, 11, 14, 11, 14, 14, 11, 11, 14, 14, 13, 11, …
$ ps_car_07_cat  <fct> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,…
$ ps_car_08_cat  <fct> 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0,…
$ ps_car_09_cat  <fct> 0, 2, 2, 3, 2, 0, 0, 2, 0, 2, 2, 0, 2, 2, 2, 0,…
$ ps_car_10_cat  <fct> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,…
$ ps_car_11_cat  <fct> 12, 19, 60, 104, 82, 104, 99, 30, 68, 104, 20, …
$ ps_car_11      <dbl> 2, 3, 1, 1, 3, 2, 2, 3, 3, 2, 3, 3, 3, 3, 1, 2,…
$ ps_car_12      <dbl> 0.4000000, 0.3162278, 0.3162278, 0.3741657, 0.3…
$ ps_car_13      <dbl> 0.8836789, 0.6188165, 0.6415857, 0.5429488, 0.5…
$ ps_car_14      <dbl> 0.3708099, 0.3887158, 0.3472751, 0.2949576, 0.3…
$ ps_car_15      <dbl> 3.605551, 2.449490, 3.316625, 2.000000, 2.00000…
$ ps_calc_01     <dbl> 0.6, 0.3, 0.5, 0.6, 0.4, 0.7, 0.2, 0.1, 0.9, 0.…
$ ps_calc_02     <dbl> 0.5, 0.1, 0.7, 0.9, 0.6, 0.8, 0.6, 0.5, 0.8, 0.…
$ ps_calc_03     <dbl> 0.2, 0.3, 0.1, 0.1, 0.0, 0.4, 0.5, 0.1, 0.6, 0.…
$ ps_calc_04     <dbl> 3, 2, 2, 2, 2, 3, 2, 1, 3, 2, 2, 2, 4, 2, 3, 2,…
$ ps_calc_05     <dbl> 1, 1, 2, 4, 2, 1, 2, 2, 1, 2, 3, 2, 1, 1, 1, 1,…
$ ps_calc_06     <dbl> 10, 9, 9, 7, 6, 8, 8, 7, 7, 8, 8, 8, 8, 10, 8, …
$ ps_calc_07     <dbl> 1, 5, 1, 1, 3, 2, 1, 1, 3, 2, 2, 2, 4, 1, 2, 5,…
$ ps_calc_08     <dbl> 10, 8, 8, 8, 10, 11, 8, 6, 9, 9, 9, 10, 11, 8, …
$ ps_calc_09     <dbl> 1, 1, 2, 4, 2, 3, 3, 1, 4, 1, 4, 1, 1, 3, 3, 2,…
$ ps_calc_10     <dbl> 5, 7, 7, 2, 12, 8, 10, 13, 11, 11, 7, 8, 9, 8, …
$ ps_calc_11     <dbl> 9, 3, 4, 2, 3, 4, 3, 7, 4, 3, 6, 9, 6, 2, 4, 5,…
$ ps_calc_12     <dbl> 1, 1, 2, 2, 1, 2, 0, 1, 2, 5, 3, 2, 3, 0, 1, 2,…
$ ps_calc_13     <dbl> 5, 1, 7, 4, 1, 0, 0, 3, 1, 0, 3, 1, 3, 4, 3, 6,…
$ ps_calc_14     <dbl> 8, 9, 7, 9, 3, 9, 10, 6, 5, 6, 6, 10, 8, 3, 9, …
$ ps_calc_15_bin <int> 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,…
$ ps_calc_16_bin <int> 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 1,…
$ ps_calc_17_bin <int> 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 1, 0, 1,…
$ ps_calc_18_bin <int> 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0,…
$ ps_calc_19_bin <int> 0, 1, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 1, 1, 1,…
$ ps_calc_20_bin <int> 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 0,…

We break up off 25% for validation.

# train-test break up
id_training <- pattern.int(nrow(porto), dimension = 0.75*nrow(porto))

x_train <- porto[id_training,] %>% choose(-goal)
x_test <- porto[-id_training,] %>% choose(-goal)
y_train <- porto[id_training, "target"]
y_test <- porto[-id_training, "target"] 

The one factor we wish to do to the information earlier than defining the community is scaling the numeric options. Binary and categorical options can keep as is, with the minor correction that for the specific ones, we’ll truly go the community the numeric illustration of the issue information.

Right here is the scaling.

train_means <- colMeans(x_train[sapply(x_train, is.double)]) %>% unname()
train_sds <- apply(x_train[sapply(x_train, is.double)], 2, sd)  %>% unname()
train_sds[train_sds == 0] <- 0.000001

x_train[sapply(x_train, is.double)] <- sweep(
  x_train[sapply(x_train, is.double)],
  2,
  train_means
  ) %>%
  sweep(2, train_sds, "/")
x_test[sapply(x_test, is.double)] <- sweep(
  x_test[sapply(x_test, is.double)],
  2,
  train_means
  ) %>%
  sweep(2, train_sds, "/")

When constructing the community, we have to specify the enter and output dimensionalities for the embedding layers. Enter dimensionality refers back to the variety of totally different symbols that “are available”; in NLP duties this is able to be the vocabulary dimension whereas right here, it’s merely the variety of values a variable can take. Output dimensionality, the capability of the interior illustration, can then be calculated based mostly on some heuristic. Under, we’ll comply with a preferred rule of thumb that takes the sq. root of the dimensionality of the enter.

In order half one of many community, right here we construct up the embedding layers in a loop, every wired to the enter layer that feeds it:

# variety of ranges per issue, required to specify enter dimensionality for
# the embedding layers
n_levels_in <- map(x_train %>% select_if(is.issue), compose(size, ranges)) %>%
  unlist() 

# output dimensionality for the embedding layers, want +1 as a result of Python is 0-based
n_levels_out <- n_levels_in %>% sqrt() %>% trunc() %>% `+`(1)

# every embedding layer will get its personal enter layer
cat_inputs <- map(n_levels_in, operate(l) layer_input(form = 1)) %>%
  unname()

# assemble the embedding layers, connecting every to its enter
embedding_layers <- vector(mode = "listing", size = size(cat_inputs))
for (i in 1:size(cat_inputs)) {
  embedding_layer <-  cat_inputs[[i]] %>% 
    layer_embedding(input_dim = n_levels_in[[i]] + 1, output_dim = n_levels_out[[i]]) %>%
    layer_flatten()
  embedding_layers[[i]] <- embedding_layer
}

In case you had been questioning concerning the flatten layer following every embedding: We have to squeeze out the third dimension (launched by the embedding layers) from the tensors, successfully rendering them rank-2. That’s as a result of we wish to mix them with the rank-2 tensor popping out of the dense layer processing the numeric options.

So as to have the ability to mix it with something, we now have to really assemble that dense layer first. It will likely be related to a single enter layer, of form 43, that takes within the numeric options we scaled in addition to the binary options we left untouched:

# create a single enter and a dense layer for the numeric information
quant_input <- layer_input(form = 43)
  
quant_dense <- quant_input %>% layer_dense(items = 64)

Are elements assembled, we wire them collectively utilizing layer_concatenate, and we’re good to name keras_model to create the ultimate graph.

intermediate_layers <- listing(embedding_layers, listing(quant_dense)) %>% flatten()
inputs <- listing(cat_inputs, listing(quant_input)) %>% flatten()

l <- 0.25

output <- layer_concatenate(intermediate_layers) %>%
  layer_dense(items = 30, activation = "relu", kernel_regularizer = regularizer_l2(l)) %>%
  layer_dropout(price = 0.25) %>%
  layer_dense(items = 10, activation = "relu", kernel_regularizer = regularizer_l2(l)) %>%
  layer_dropout(price = 0.25) %>%
  layer_dense(items = 5, activation = "relu", kernel_regularizer = regularizer_l2(l)) %>%
  layer_dropout(price = 0.25) %>%
  layer_dense(items = 1, activation = "sigmoid", kernel_regularizer = regularizer_l2(l))

mannequin <- keras_model(inputs, output)

Now, should you’ve truly learn by the entire of this half, you might want for a neater approach to get up to now. So let’s swap to function specs for the remainder of this publish.

Function specs to the rescue

In spirit, the way in which function specs are outlined follows the instance of the recipes package deal. (It gained’t make you hungry, although.) You initialize a function spec with the prediction goal – feature_spec(goal ~ .), after which use the %>% to inform it what to do with particular person columns. “What to do” right here signifies two issues:

  • First, find out how to “learn in” the info. Are they numeric or categorical, and if categorical, what am I presupposed to do with them? For instance, ought to I deal with all distinct symbols as distinct, leading to, probably, an infinite rely of classes – or ought to I constrain myself to a hard and fast variety of entities? Or hash them, even?
  • Second, elective subsequent transformations. Numeric columns could also be bucketized; categorical columns could also be embedded. Or options may very well be mixed to seize interplay.

On this publish, we show the usage of a subset of step_ capabilities. The vignettes on Function columns and Function specs illustrate extra capabilities and their software.

Ranging from the start once more, right here is the whole code for information read-in and train-test break up within the function spec model.

Information-prep-wise, recall what our objectives are: depart alone if binary; scale if numeric; embed if categorical. Specifying all of this doesn’t want various strains of code:

Be aware how right here we’re passing within the coaching set, and identical to with recipes, we gained’t have to repeat any of the steps for the validation set. Scaling is taken care of by scaler_standard(), an elective transformation operate handed in to step_numeric_column. Categorical columns are supposed to make use of the whole vocabulary and pipe their outputs into embedding layers.

Now, what truly occurred after we known as match()? Rather a lot – for us, as we removed a ton of handbook preparation. For TensorFlow, nothing actually – it simply got here to learn about just a few items within the graph we’ll ask it to assemble.

However wait, – don’t we nonetheless must construct up that graph ourselves, connecting and concatenating layers? Concretely, above, we needed to:

  • create the proper variety of enter layers, of right form; and
  • wire them to their matching embedding layers, of right dimensionality.

So right here comes the actual magic, and it has two steps.

First, we simply create the enter layers by calling layer_input_from_dataset:

`

inputs <- layer_input_from_dataset(porto %>% choose(-goal))

And second, we are able to extract the options from the function spec and have layer_dense_features create the required layers based mostly on that info:

layer_dense_features(ft_spec$dense_features())

With out additional ado, we add just a few dense layers, and there may be our mannequin. Magic!

output <- inputs %>%
  layer_dense_features(ft_spec$dense_features()) %>%
  layer_dense(items = 30, activation = "relu", kernel_regularizer = regularizer_l2(l)) %>%
  layer_dropout(price = 0.25) %>%
  layer_dense(items = 10, activation = "relu", kernel_regularizer = regularizer_l2(l)) %>%
  layer_dropout(price = 0.25) %>%
  layer_dense(items = 5, activation = "relu", kernel_regularizer = regularizer_l2(l)) %>%
  layer_dropout(price = 0.25) %>%
  layer_dense(items = 1, activation = "sigmoid", kernel_regularizer = regularizer_l2(l))

mannequin <- keras_model(inputs, output)

How can we feed this mannequin? Within the non-feature-columns instance, we’d have needed to feed every enter individually, passing a listing of tensors. Now we are able to simply go it the whole coaching set all of sudden:

mannequin %>% match(x = coaching, y = coaching$goal)

Within the Kaggle competitors, submissions are evaluated utilizing the normalized Gini coefficient, which we are able to calculate with the assistance of a brand new metric out there in Keras, tf$keras$metrics$AUC(). For coaching, we are able to use an approximation to the AUC as a consequence of Yan et al. (2003) (Yan et al. 2003). Then coaching is as simple as:

auc <- tf$keras$metrics$AUC()

gini <- custom_metric(identify = "gini", operate(y_true, y_pred) {
  2*auc(y_true, y_pred) - 1
})

# Yan, L., Dodier, R., Mozer, M. C., & Wolniewicz, R. (2003). 
# Optimizing Classifier Efficiency through an Approximation to the Wilcoxon-Mann-Whitney Statistic.
roc_auc_score <- operate(y_true, y_pred) {

  pos = tf$boolean_mask(y_pred, tf$forged(y_true, tf$bool))
  neg = tf$boolean_mask(y_pred, !tf$forged(y_true, tf$bool))

  pos = tf$expand_dims(pos, 0L)
  neg = tf$expand_dims(neg, 1L)

  # authentic paper suggests efficiency is strong to actual parameter alternative
  gamma = 0.2
  p     = 3

  distinction = tf$zeros_like(pos * neg) + pos - neg - gamma

  masked = tf$boolean_mask(distinction, distinction < 0.0)

  tf$reduce_sum(tf$pow(-masked, p))
}

mannequin %>%
  compile(
    loss = roc_auc_score,
    optimizer = optimizer_adam(),
    metrics = listing(auc, gini)
  )

mannequin %>%
  match(
    x = coaching,
    y = coaching$goal,
    epochs = 50,
    validation_data = listing(testing, testing$goal),
    batch_size = 512
  )

predictions <- predict(mannequin, testing)
Metrics::auc(testing$goal, predictions)

After 50 epochs, we obtain an AUC of 0.64 on the validation set, or equivalently, a Gini coefficient of 0.27. Not a foul outcome for a easy absolutely related community!

We’ve seen how utilizing function columns automates away a variety of steps in organising the community, so we are able to spend extra time on truly tuning it. That is most impressively demonstrated on a dataset like this, with greater than a handful categorical columns. Nevertheless, to clarify a bit extra what to concentrate to when utilizing function columns, it’s higher to decide on a smaller instance the place we are able to simply do some peeking round.

Let’s transfer on to the second software.

Interactions, and what to look out for

To show the usage of step_crossed_column to seize interactions, we make use of the rugged dataset from Richard McElreath’s rethinking package deal.

We wish to predict log GDP based mostly on terrain ruggedness, for a variety of international locations (170, to be exact). Nevertheless, the impact of ruggedness is totally different in Africa versus different continents. Citing from Statistical Rethinking

It is sensible that ruggedness is related to poorer international locations, in many of the world. Rugged terrain means transport is troublesome. Which suggests market entry is hampered. Which suggests lowered gross home product. So the reversed relationship inside Africa is puzzling. Why ought to troublesome terrain be related to larger GDP per capita?

If this relationship is in any respect causal, it might be as a result of rugged areas of Africa had been protected towards the Atlantic and Indian Ocean slave trades. Slavers most well-liked to raid simply accessed settlements, with straightforward routes to the ocean. These areas that suffered below the slave commerce understandably proceed to undergo economically, lengthy after the decline of slave-trading markets. Nevertheless, an end result like GDP has many influences, and is moreover a wierd measure of financial exercise. So it’s onerous to make certain what’s occurring right here.

Whereas the causal state of affairs is troublesome, the purely technical one is well described: We wish to study an interplay. We might depend on the community discovering out by itself (on this case it most likely will, if we simply give it sufficient parameters). However it’s a wonderful event to showcase the brand new step_crossed_column.

Loading the dataset, zooming in on the variables of curiosity, and normalizing them the way in which it’s achieved in Rethinking, we now have:

Observations: 170
Variables: 3
$ log_gdp <dbl> 0.8797119, 0.9647547, 1.1662705, 1.1044854, 0.9149038,…
$ rugged  <dbl> 0.1383424702, 0.5525636891, 0.1239922606, 0.1249596904…
$ africa  <int> 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, …

Now, let’s first overlook concerning the interplay and do the very minimal factor required to work with this information. rugged must be a numeric column, whereas africa is categorical in nature, which implies we use one of many step_categorical_[...] capabilities on it. (On this case we occur to know there are simply two classes, Africa and not-Africa, so we might as effectively deal with the column as numeric like within the earlier instance; however in different functions that gained’t be the case, so right here we present a technique that generalizes to categorical options basically.)

So we begin out making a function spec and including the 2 predictor columns. We test the outcome utilizing feature_spec’s dense_features() technique:

ft_spec <- coaching %>%
  feature_spec(log_gdp ~ .) %>%
  step_numeric_column(rugged) %>%
  step_categorical_column_with_identity(africa, num_buckets = 2) 
  match()

ft_spec$dense_features()
$rugged
NumericColumn(key='rugged', form=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None)

Hm, that doesn’t look too good. The place’d africa go? In actual fact, there may be another factor we must always have achieved: convert the specific column to an indicator column. Why?

The rule of thumb is, at any time when you’ve one thing categorical, together with crossed, you might want to then remodel it into one thing numeric, which incorporates indicator and embedding.

Being a heuristic, this rule works general, and it matches our instinct. There’s one exception although, step_bucketized_column, which though it “feels” categorical truly doesn’t want that conversion.

Subsequently, it’s best to complement that instinct with a easy lookup diagram, which can be a part of the function columns vignette.

With this diagram, the easy rule is: We all the time want to finish up with one thing that inherits from DenseColumn. So:

  • step_numeric_column, step_indicator_column, and step_embedding_column are standalone;
  • step_bucketized_column is, too, nonetheless categorical it “feels”; and
  • all step_categorical_column_[...], in addition to step_crossed_column, have to be remodeled utilizing one the dense column varieties.

Determine 1: To be used with Keras, all options want to finish up inheriting from DenseColumn by some means.

Thus, we are able to repair the state of affairs like so:

ft_spec <- coaching %>%
  feature_spec(log_gdp ~ .) %>%
  step_numeric_column(rugged) %>%
  step_categorical_column_with_identity(africa, num_buckets = 2) %>%
  step_indicator_column(africa) %>%
  match()

and now ft_spec$dense_features() will present us

$rugged
NumericColumn(key='rugged', form=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None)

$indicator_africa
IndicatorColumn(categorical_column=IdentityCategoricalColumn(key='africa', number_buckets=2.0, default_value=None))

What we actually wished to do is seize the interplay between ruggedness and continent. To this finish, we first bucketize rugged, after which cross it with – already binary – africa. As per the foundations, we lastly remodel into an indicator column:

ft_spec <- coaching %>%
  feature_spec(log_gdp ~ .) %>%
  step_numeric_column(rugged) %>%
  step_categorical_column_with_identity(africa, num_buckets = 2) %>%
  step_indicator_column(africa) %>%
  step_bucketized_column(rugged,
                         boundaries = c(0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.8)) %>%
  step_crossed_column(africa_rugged_interact = c(africa, bucketized_rugged),
                      hash_bucket_size = 16) %>%
  step_indicator_column(africa_rugged_interact) %>%
  match()

Taking a look at this code you might be asking your self, now what number of options do I’ve within the mannequin? Let’s test.

$rugged
NumericColumn(key='rugged', form=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None)

$indicator_africa
IndicatorColumn(categorical_column=IdentityCategoricalColumn(key='africa', number_buckets=2.0, default_value=None))

$bucketized_rugged
BucketizedColumn(source_column=NumericColumn(key='rugged', form=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), boundaries=(0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.8))

$indicator_africa_rugged_interact
IndicatorColumn(categorical_column=CrossedColumn(keys=(IdentityCategoricalColumn(key='africa', number_buckets=2.0, default_value=None), BucketizedColumn(source_column=NumericColumn(key='rugged', form=(1,), default_value=None, dtype=tf.float32, normalizer_fn=None), boundaries=(0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.8))), hash_bucket_size=16.0, hash_key=None))

We see that each one options, authentic or remodeled, are stored, so long as they inherit from DenseColumn. Because of this, for instance, the non-bucketized, steady values of rugged are used as effectively.

Now organising the coaching goes as anticipated.

inputs <- layer_input_from_dataset(df %>% choose(-log_gdp))

output <- inputs %>%
  layer_dense_features(ft_spec$dense_features()) %>%
  layer_dense(items = 8, activation = "relu") %>%
  layer_dense(items = 8, activation = "relu") %>%
  layer_dense(items = 1)

mannequin <- keras_model(inputs, output)

mannequin %>% compile(loss = "mse", optimizer = "adam", metrics = "mse")

historical past <- mannequin %>% match(
  x = coaching,
  y = coaching$log_gdp,
  validation_data = listing(testing, testing$log_gdp),
  epochs = 100)

Simply as a sanity test, the ultimate loss on the validation set for this code was ~ 0.014. However actually this instance did serve totally different functions.

In a nutshell

Function specs are a handy, elegant manner of constructing categorical information out there to Keras, in addition to to chain helpful transformations like bucketizing and creating crossed columns. The time you save information wrangling might go into tuning and experimentation. Get pleasure from, and thanks for studying!

Yan, Lian, Robert H Dodier, Michael Mozer, and Richard H Wolniewicz. 2003. “Optimizing Classifier Efficiency through an Approximation to the Wilcoxon-Mann-Whitney Statistic.” In Proceedings of the twentieth Worldwide Convention on Machine Studying (ICML-03), 848–55.