Building Token Recommender in Google Cloud Platform

In this article I will guide you through the process of creating an ERC20 token recommendation system built with TensorFlow, Cloud Machine Learning Engine, Cloud Endpoints, and App Engine. The solution is based on the tutorial article by Google. The data used for training the recommendation system is taken from our public Ethereum dataset in BigQuery.

The article is broken down into the following parts:

  1. Intro to collaborative filtering for recommendation systems.
  2. Creating and training the model for token recommendation system.
  3. Tuning hyperparameters in Cloud ML Engine.
  4. Deploying the recommendation system to Cloud Endpoints and App Engine.

Intro to collaborative filtering for recommendation systems

The collaborative filtering technique is a powerful method for generating user recommendations. Collaborative filtering relies only on observed user behavior to make recommendations — no profile data or content access is necessary.

The technique is based on the following observations:

  • Users who interact with items in a similar manner (for example, buying the same tokens or viewing the same articles) share one or more hidden preferences.
  • Users with shared preferences are likely to respond in the same way to the same items.

The collaborative filtering problem can be solved using matrix factorization. Suppose you have a matrix consisting of user IDs and their interactions with your products. Each row corresponds to a unique user, and each column corresponds to an item. The item could be a product in a catalog, an article, or a token. Each entry in the matrix captures a user’s rating or preference for a single item. The rating could be explicit, directly generated by user feedback, or it could be implicit, based on user purchases or the number of interactions with an article or a token.

Ratings matrix

The matrix factorization method assumes that there is a set of attributes common to all items, with items differing in the degree to which they express these attributes. Furthermore, the matrix factorization method assumes that each user has their own expression for each of these attributes, independent of the items. In this way, a user’s item rating can be approximated by summing the user’s strength for each attribute weighted by the degree to which the item expresses this attribute. These attributes are sometimes called hidden or latent factors.

To translate the existence of latent factors into the matrix of ratings, you do this: for a set of users U of size u and items I of size i, you pick an arbitrary number k of latent factors and factorize the large matrix R into two much smaller matrices X (the “row factor”) and Y (the “column factor”). Matrix X has dimension u × k, and Y has dimension k × i.

Approximating the ratings matrix with row and column factors

To calculate the rating of user u for item i, you take the dot product of the two vectors. The loss function can be defined as Root-mean-square error (RMSE) between the actual rating and the rating calculated from the latent factors.

For our token recommender we will use the percentage of supply a user is holding for a particular token, as the implicit rating in the user rating matrix.

Creating and training the model for token recommendation system

Check out the code and install the dependencies:

wget Miniconda2-latest-MacOSX-x86_64.shgit clone --branch medium_article
cd token-recommender
conda create -n token_recommender
conda install -n token_recommender --file conda.txt
source activate token_recommender
pip install -r requirements.txt
pip install tensorflow==1.4.1

Query token ratings from BigQuery:

Run the following query in BigQuery and export the results to a GCS bucket

with top_tokens as (
select token_address, count(1) as transfer_count
from `bigquery-public-data.crypto_ethereum.token_transfers` as token_transfers
group by token_address
order by transfer_count desc
limit 1000
token_balances as (
with double_entry_book as (
select token_address, to_address as address, cast(value as float64) as value, block_timestamp
from `bigquery-public-data.crypto_ethereum.token_transfers`
union all
token_address, from_address as address, -cast(value as float64) as value, block_timestamp
from `bigquery-public-data.crypto_ethereum.token_transfers`
select double_entry_book.token_address, address, sum(value) as balance
from double_entry_book
join top_tokens on top_tokens.token_address = double_entry_book.token_address
where address != '0x0000000000000000000000000000000000000000'
group by
token_address, address
having balance > 0
token_supplies as (
select token_address, sum(balance) as supply
from token_balances
group by token_address
token_balances.address as user_address,
balance/supply * 100 as rating
from token_balances
join token_supplies on token_supplies.token_address = token_balances.token_address
where balance/supply * 100 > 0.001

The above SQL queries top 1000 tokens by transfers count, calculates the balances for each token, and outputs (token_address, user_address, rating) triples. Rating there is calculated as the percentage of supply held by the user. This filter — where balance/supply * 100 > 0.001 — prevents airdrops appearing in the result.

Understand the code structure

The model code is contained in the wals_ml_engine directory. The code's high-level functionality is implemented by the following files:

  1. — launches various types of Cloud Machine Learning Engine jobs.
  2. — parses the arguments for the Cloud Machine Learning Engine job and executes training.
  3. — loads the dataset; creates two sparse matrices from the data, one for training and one for testing; executes WALS on the training sparse matrix of ratings.
  4.— creates the WALS model; executes the WALS algorithm; calculates the root-mean-square error (RMSE) for a set of row/column factors and a ratings matrix.

The csv file is loaded in the file:

headers = ['token_address', 'user_address', 'rating']
balances_df = pd.read_csv(input_file,
'token_address': np.str,
'user_address': np.str,
'rating': np.float32,

Then the following arrays are created:

  • an array of user addresses,
  • an array of token addresses,
  • an array of triplets with 0-based user and token indexes and the corresponding ratings.

These triplets are then randomly split into test and train datasets and converted to sparse matrices:

test_set_size = len(ratings) / TEST_SET_RATIO
test_set_idx = np.random.choice(xrange(len(ratings)),
size=test_set_size, replace=False)
test_set_idx = sorted(test_set_idx)

# sift ratings into train and test sets
ts_ratings = ratings[test_set_idx]
tr_ratings = np.delete(ratings, test_set_idx, axis=0)
# create training and test matrices as coo_matrix's
u_tr, i_tr, r_tr = zip(*tr_ratings)
tr_sparse = coo_matrix((r_tr, (u_tr, i_tr)), shape=(n_users, n_items))

u_ts, i_ts, r_ts = zip(*ts_ratings)
test_sparse = coo_matrix((r_ts, (u_ts, i_ts)), shape=(n_users, n_items))

The WALS model is created in wals_model method in, and the factorization is done in simple_train method in the same file. The result are the row and column factors in numpy format.

Train the model locally and in Google ML Engine

To train the model locally run the following command, specifying the path to the CSV file exported on the previous step:

./ local gs://your_bucket/data/token_balances.csv

The output should look like the following:

INFO:tensorflow:Train Start: 2019-01-10 23:22:06
INFO:tensorflow:Train Finish: 2019-01-10 23:22:12
INFO:tensorflow:train RMSE = 0.76
INFO:tensorflow:test RMSE = 0.95

The RMSE corresponds to the average error in the predicted ratings compared to the test set. On average, each rating produced by the algorithm is within ± 0.95 percentage points of the actual user rating in the test set. The WALS algorithm performs much better with tuned hyperparameters, as shown in the following section.

To run it in Cloud ML Engine:

./ train gs://your_bucket/data/token_balances.csv

You can monitor the status and output of the job on the Jobs page of the ML Engine section of the GCP Console. Click Logs to view the job output.

After factorization, the factor matrices are saved in four separate files in numpy format so they can be used to perform recommendations:

  • user.npy —an array of user addresses used for mapping user indexes to user addresses
  • item.npy — an array of token addresses used for mapping token indexes to token addresses
  • row.npy — users latent factors
  • col.npy — tokens latent factors

When training locally, you can find those files under wals_ml_engine/jobs directory.

To test out the recommendations use the following code:

import numpy as np
from model import generate_recommendations

user_address = '0x8c373ed467f3eabefd8633b52f4e1b2df00c9fe8'
already_rated = ['0x006bea43baa3f7a6f765f14f10a1a1b08334ef45','0x5102791ca02fc3595398400bfe0e33d7b6c82267','0x68d57c9a1c35f63e2c83ee8e49a64e9d70528d25','0xc528c28fec0a90c083328bc45f587ee215760a0f']
k = 5

model_dir = '../jobs/wals_ml_local_20190107_235006'

user_map = np.load(model_dir + "/model/user.npy")
item_map = np.load(model_dir + "/model/item.npy")
row_factor = np.load(model_dir + "/model/row.npy")
col_factor = np.load(model_dir + "/model/col.npy")
user_idx = np.searchsorted(user_map, user_address)
user_rated = [np.searchsorted(item_map, i) for i in already_rated]

recommendations = generate_recommendations(user_idx, user_rated, row_factor, col_factor, k)

tokens = [item_map[i] for i in recommendations]


Tuning hyperparameters in Cloud ML Engine

You can find the configuration file for hyperparameters tuning here.

  • latent_factors — the number of latent factors (min:5, max:50).
  • regularization — L2 Regularization constant (min: 0.001, max: 10.0).
  • unobs_weight — unobserved weight (min: 0.001, max: 5.0).
  • feature_wt_exp — feature weight exponent (min: 0.0001, max: 10).
  • num_iters — number of alternating least squares iterations (min: 10, max 20)

To tune the hyperparameters, first change the BUCKET variable in to your bucket. Then run the following command:

./ tune gs://your_bucket/data/token_balances.csv

You can see the progress of tuning in Cloud ML Engine console. The results of hyperparameter tuning are stored in the Cloud ML Engine job data, which you can access in the Jobs page. The job results include the best RMSE score across all trials of the summary metric.

Below are the best parameters from my tuning, which you can also find in the repository:

  • latent_factors — 22
  • regularization — 0.12
  • unobs_weight — 0.001
  • feature_wt_exp —9.43
  • num_iters — 20

The error though is just slightly lower comparing to the default parameters:

INFO:tensorflow:train RMSE = 0.97
INFO:tensorflow:test RMSE = 0.87

Deploying the recommendation system to Google App Engine

You can find REST API definition in Swagger format for serving token recommendations in the repository: openapi.yaml. The implementation of the API for App Engine is in the file.

First, prepare to deploy the API endpoint service:

cd scripts

The output of this command should look like the following:

To deploy:  gcloud endpoints services deploy /var/folders/t0/y38g0z2s6jqcnwp8452j026h0000gp/T/tmp.fIelYqSh8B.yaml

Run the provided command:

gcloud endpoints services deploy /var/folders/t0/y38g0z2s6jqcnwp8452j026h0000gp/T/tmp.fIelYqSh8B.yaml

Create the bucket where the app will read the model from:

export BUCKET=gs://recserve_$(gcloud config get-value project 2> /dev/null)
gsutil mb ${BUCKET}

Upload the token_balances.csv file to the bucket:

gsutil cp ../data/token_balances.csv ${BUCKET}/data/

Train the model and upload the model files to the bucket:

./ local ${BUCKET}/data/token_balances.csv \
--use-optimized --output-dir ${BUCKET}

Create an App Engine application:

gcloud app create --region=us-central1
gcloud app update --no-split-health-checks

Prepare to deploy the App Engine application:


The following output appears:

To deploy:  gcloud -q app deploy ../app/app_template.yaml_deploy.yaml

Run the provided command:

gcloud -q app deploy ../app/app_template.yaml_deploy.yaml

After the app is deployed you will be able to query the api: https://${project_id} (replace ${project_id} with your value)

token_addresses: [

You can try out the recommendations at Read our article where we describe how we evaluated and improved the recommendation system:




A collection of technical articles and blogs published or curated by Google Cloud Developer Advocates. The views expressed are those of the authors and don't necessarily reflect those of Google.

Recommended from Medium

MPST: Movie Plot Synopses with Tags

Movie Tags

Driverless Cars-Why the World isn’t Ready

Image-based ML techniques to classify billions of e-commerce products into thousands of categories

BERT for dummies —  Step by Step Tutorial

AI-900: Exam Questions (2/7)

Nvidia Transfer Learning Toolkit — A Comprehensive Guide

Machine Learning and the Production Gap

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Evgeny Medvedev

Evgeny Medvedev

Creator of, Co-founder of and, Google Cloud GDE, AWS Certified Solutions Architect

More from Medium

Trillo Workbench for Similarity Matching on the Google Cloud

Codeq’s Summarizer Updated with Summary Length Option

Using Google Maps Places API to “validate” locations in a conversation with a virtual agent

Measuring and Improving Speech-to-Text Accuracy | Google Cloud Platform