SCD Implementation with Databricks Delta

Slowly Changing Dimensions (SCD) are the most commonly used advanced dimensional technique used in dimensional data warehouses. Slowly changing dimensions are used when you wish to capture the data changes (CDC) within the dimension over time. Two typical SCD scenarios: SCD Type 1 and SCD Type 2. Type 1 – For this type of slowly changing dimension you simply overwrite the existing data values with new data values. Type 2 – This is the most commonly used type of slowly changing dimension. For this type of slowly changing dimension, add a new record encompassing the change and mark the old record as inactive.  In this blog, we are going to describe how we implement SCD Type 1 and SCD Type 2 with Azure Databricks. SCD Type 1&2 are newly supported by Databricks Delta. Please see the office document link for the command.

Data Preparation

Original Records: target.csv

G87D744D-345T-46AD-BD9D-B18CB66345YTProduct0user0@dummy.comProduct0 Desc
D87D7FFD-E03B-46AD-BD9D-B18CB6632DC1Product1user1@dummy.comProduct1 Desc
FF7D7FFD-E03B-46AD-BD9D-B18CB6632WW3Product2user2@dummy.comProduct2 Desc
TT7D7FFD-E03B-46AD-BD9D-B18CB6632256Product3user3@dummy.comProduct3 Desc

Update Records: source.csv

D87D7FFD-E03B-46AD-BD9D-B18CB6632DC1Product1user1@dummy.comProduct1 Description Changed Only
FF7D7FFD-E03B-46AD-BD9D-B18CB6632WW3Product2userchanged@dummy.comProduct2 Owner and Desciption Changed
TT7D7FFD-E03B-46AD-BD9D-B18CB6632256Product3user3@dummy.comProduct3 Description
2A6CE7F2-4C6F-41DF-9819-235021DC1226Product4user4@dummy.comNew Product

ID: Business Key column; Name: SCD Type 2 column; Owner: SCD Type 2 column; Descriptio: SCD Type 1 column.

Expected result:

Product0 row will remain the same; Product1 row only description will be updated; Original Product2 row will be inactive, and new row of Product2 is added; Product3 will remain the same; Product4 new row will be added.

Data Load and Transformation

Upload the targer.csv and source.csv into cloud, e.g. Azure blob storage or DBFS. Then load the two data files into two DataFrames.

# Load Data
target_df ="sep", ",").option("header", "true").option("inferSchema", "true").csv("/mnt/vpa-raw-data-dev/POC/target.csv")
source_df ="sep", ",").option("header", "true").option("inferSchema", "true").csv("/mnt/vpa-raw-data-dev/POC/source.csv")

In order to use MERGE command, we need save to two DataFrames into Delta table. More info about Delta table, please see the link here.

# Delta tables
spark.sql("DROP TABLE IF EXISTS source")
spark.sql("CREATE TABLE source (ID STRING, Name STRING, Owner STRING, Description STRING) USING DELTA LOCATION '/mnt/vpa-raw-data-dev/POC/source'")
spark.sql("TRUNCATE TABLE source")
source_df.write.format("delta").mode("overwrite").option("mergeSchema", "true").save('/mnt/vpa-raw-data-dev/POC/source')

from pyspark.sql.functions import *
from pyspark.sql.types import *
spark.sql("DROP TABLE IF EXISTS target")
spark.sql("CREATE TABLE target (ID STRING, Name STRING, Owner STRING, Description STRING, RowStatus STRING) USING DELTA LOCATION '/mnt/vpa-raw-data-dev/POC/target'")
spark.sql("TRUNCATE TABLE target")
target_df = target_df.withColumn('RowStatus', lit(None).cast(StringType()))

SCD Merge

We have to use two MERGE command to cover the four scenarios: SCD 1, SCD 2, New rows, and unchanged rows, as one MERGE command only support one UPDATE/INSERT within.

-- Update SCD Type 2 rows (RowStatus = 2) and Insert Not Match rows (RowStatus = 3)
USING source
ON target.ID = source.ID
WHEN MATCHED AND (source.Owner <> target.Owner OR source.Name <> target.Name) THEN
    target.RowStatus = 2,
    target.Owner = source.Owner,
    target.Name = source.Name,
    target.Description = source.Description  
  THEN INSERT (ID,Name,Owner,Description,RowStatus) VALUES (source.ID,source.Name,source.Owner,source.Description,3)
-- Merge SCD Type 1 update (RowStatus = 1)
USING source
ON target.ID = source.ID
WHEN MATCHED AND (target.RowStatus IS NULL) AND (source.Description <> target.Description) THEN
    target.RowStatus = 1,
    target.Description = source.Description

Check the merge result in target table if both SCD Type 1 and 2 update as expected.

Update and Insert

When the merged data set in target table, we need use this target dataset to update the dimension tables, probably in traditional database(SQL Server/MySQL…). Here we use Azure SQL Database as an example.

Spark JDBC doesn’t support Update command. One workaround is use Spark Connector. In order to achieve this, we need create one temporary table to store those SCD Type 1 and Type 2 rows. Then update the dimension table with the temporary table through Spark Connector.

#Filter out SCD Type 1 and 2 rows from target Delta table, and save into one temp table in Azure SQL
scd12_df = spark.sql("SELECT ID, Name, Owner, Description, RowStatus FROM target WHERE ( RowStatus = 2 OR RowStatus = 1)")
scd12_df.write.mode("overwrite").jdbc(url = jdbcUrl, table = "Scd_tmp", properties = connectionProperties)
import org.apache.spark.sql.SQLContext

//Update columns value for those SCD Type 1 change only row
val scd1_query = """
              |UPDATE Scd
              |SET Scd.Description = Scd_tmp.Description
              |FROM Scd
              |INNER JOIN Scd_tmp
              |ON Scd.ID = Scd_tmp.ID AND Scd_tmp.RowStatus = '1';

val scd1_config = Config(Map(
  "url"            -> "dummy url",
  "databaseName"   -> "dummy databaseName",
  "user"           -> "dummy user",
  "password"       -> "dummy pwd",
  "queryCustom"    -> scd1_query

//Update SCD Type 2 row: Set Active_Record as 0, and Record_EndDate as current datatime.
val scd2_query2 = """
              |UPDATE Scd
              |SET Scd.Active_Record = '0', Scd.Record_EndDate = GETDATE()
              |FROM Scd
              |INNER JOIN Scd_tmp
              |ON Scd.ID = Scd_tmp.ID AND Scd_tmp.RowStatus = '2';

val scd2_config = Config(Map(
  "url"            -> "dummy url",
  "databaseName"   -> "dummy databaseName",
  "user"           -> "dummy user",
  "password"       -> "dummy pwd",
  "queryCustom"    -> scd1_query

Finally insert the those from Spark into SQL tables through JDBC API.

newinserted_df = spark.sql("SELECT ID, Name, Owner, Description FROM target WHERE RowStatus = '3'")
newinserted_df.write.mode("append").jdbc(url = jdbcUrl, table = "Scd", properties = connectionProperties)

Handling Excel Data in Azure Databricks

By now, there is no default support of loading data from Spark in Cloud.(Here we take Azure Databricks as the example). Based on research, some links sound helpful.

This article describes some practice based on our recent projects.

Data Preparation

Sample MS Excel data as below: Two sheets: Companies, and Users


Solution 1: Read Excel data using Pandas, then convert Pandas DataFrame into Spark DataFrame.

from datetime import datetime
from pyspark.sql.functions import *
from pyspark.sql.types import *
import pandas as pd

filepath = '/dbfs/xxx/xxx/Samples.xlsx'
companies_pd_df = pd.read_excel(filepath, sheet_name='Companies')
companiesSchema = StructType([
  StructField("CompanyID", IntegerType(), False),
  StructField("CompanyName", StringType(), False),
  StructField("Description", StringType(), False)])
companies_df = spark.createDataFrame(companies_pd_df, schema=companiesSchema)

Note that:

  1. While convert Pandas DataFrame into Spark DataFrame, we need to manually define the Schema, otherwise the conversion will fail probably.
  2. In some cases, the created Spark DataFrame may display some dummy data or additional unnecessary row. In this case, we may filter out those unnecessary rows. e.g. companies_Df = companies_df.filter(isnan(“CompaniesID”) != True)

Solution 2: Use Spark Excel.

More details documentation can be found here. Code sample in Azure Databricks:


import org.apache.spark.sql._
import org.apache.spark.sql.types._
import com.crealytics.spark.excel._

val companiesSchema = StructType(Array(
  StructField("CompanyID", IntegerType, nullable = true),
  StructField("CompanyName", StringType, nullable = true),
  StructField("Description", StringType, nullable = true)))
val filepath = "/mnt/xxx/xxx/Samples.xlsx"
val df =
    .option("sheetName", "Companies")
    .option("useHeader", "true")

Note that:

  1. Currently Spark Excel plugin is only available for Scala, not for Python yet. Hence we have to use magic command for Python notebook.
  2. Error/Exceptions may happens for some versions. I have tested out this successfully with version com.crealytics:spark-excel_2.11:0.12.0

Google Analytics Raw Data Ingest

Google Analytics is a very popular tool for those customers that want analytics insights for their website or apps, without building their own data analytics system or big data platform.

In some cases, Google Analytics(GA) report/dashboards or its reporting api are not matching our needs perfectly. Hence we did some research to investigate how to ingest  GA raw data.

Solution 1:  Google Analytics 360

GA 360 supports features that export session and hit data into Google BigQuery. But the biggest challenge is that GA 360 price starts from US$150000 per year. This may be one concern for small or startup companies.

Solution 2: Third party tool.



Solution 3: Custom code to implement export ga raw into Bigquery. Some examples found.

Inspired by the above articles, we are able to send the raw hit level data to GA and another destination (e.g. BigQuery) at the same time. Then we can continuously copy data from BigQuery into Blob Storage through Azure Data Factory. The data flow is as below:

Kaggle AllState Competition in Azure Databricks

This post, we will describe how to practice one Kaggle competition process with Azure Databricks. Compared to run our training and tuning phase in local machines or single servers, it is quite fast that we can train our model in Azure Databricks with Spark.

Kaggle Allstate Claims Severity

When you’ve been devastated by a serious car accident, your focus is on the things that matter the most: family, friends, and other loved ones. Pushing paper with your insurance agent is the last place you want your time or mental energy spent. This is why Allstate, a personal insurer in the United States, is continually seeking fresh ideas to improve their claims service for the over 16 million households they protect.

Allstate is currently developing automated methods of predicting the cost, and hence severity, of claims. In this recruitment challenge, Kagglers are invited to show off their creativity and flex their technical chops by creating an algorithm which accurately predicts claims severity. Aspiring competitors will demonstrate insight into better ways to predict claims severity for the chance to be part of Allstate’s efforts to ensure a worry-free customer experience.

Data and Preparation

Download the data from Upload the data file into DBFS or Azure blob storage, then read train data and test data into DataFrame.

import re
from pyspark.sql import SparkSession
from import StringIndexer, VectorAssembler
from import RandomForestRegressor, RandomForestRegressionModel
from import Pipeline, PipelineModel
from import RegressionEvaluator
from import ParamGridBuilder, CrossValidator
from pyspark.mllib.evaluation import RegressionMetrics
print("Read and load data started...")
trainInput = (
.option("header", "true")
.option("inferSchema", "true")

testInput = (
.option("header", "true")
.option("inferSchema", "true")
print("Read and load data completed...")
data = trainInput.withColumnRenamed("loss", "label")
[trainingData, validationData] = data.randomSplit([0.7, 0.3])
testData = testInput.cache()

Please note that you should replace the above bdfs path with your one. For Spark Dataframe/Dataset/RDD, better we cache them for future release. Here we cache trainingData, validationData and testData.

Modeling and Training

print("Feature engineering...")
print("Handle categories data...")
# Use StringIndexer or OneHotEncoder for categories columns
isCateg = lambda c: c.startswith("cat")
categNewCol = lambda c: "idx_{0}".format(c) if (isCateg(c)) else c

stringIndexerStages = map(lambda c: StringIndexer(inputCol=c, outputCol=categNewCol(c))
.fit(, filter(isCateg, trainingData.columns))

removeTooManyCategs = lambda c: not re.match(r"cat(109$|110$|112$|113$|116$)", c)

# Keep those feature columns only
onlyFeatureCols = lambda c: not re.match(r"id|label", c)

featureCols = map(categNewCol, 

# Assemble features
assembler = VectorAssembler(inputCols=featureCols, outputCol="features")
print("Features generation and assembly completed...")
print("Building Random Forest for regression..")
algo = RandomForestRegressor(featuresCol="features", labelCol="label")

stages = stringIndexerStages

#Build pipeline
pipeline = Pipeline(stages=stages)
print("K fold cross validation...")
numTrees = [5, 20]
maxDepth = [4, 6]
maxBins = [32]
numFolds = 3

paramGrid = (ParamGridBuilder()
.addGrid(algo.numTrees, numTrees)
.addGrid(algo.maxDepth, maxDepth)
.addGrid(algo.maxBins, maxBins)

cv = CrossValidator(estimator=pipeline,

cvModel =

Results Metrics and Prediction

trainPredictionsAndLabels = cvModel.transform(trainingData).select("label", "prediction").rdd

validPredictionsAndLabels = cvModel.transform(validationData).select("label", "prediction").rdd

trainRegressionMetrics = RegressionMetrics(trainPredictionsAndLabels)
validRegressionMetrics = RegressionMetrics(validPredictionsAndLabels)

bestModel = cvModel.bestModel
featureImportances = bestModel.stages[-1].featureImportances.toArray()

print("TrainingData count: {0}".format(trainingData.count()))
print("ValidationData count: {0}".format(validationData.count()))
print("TestData count: {0}".format(testData.count()))
print("Param algoNumTrees = {0}".format(",".join(map(lambda x:str(x), numTrees))))
print("Param algoMaxDepth = {0}".format(",".join(map(lambda x:str(x), maxDepth))))
print("Param algoMaxBins = {0}".format(",".join(map(lambda x:str(x), maxBins))))
print("Param numFolds = {0}".format(numFolds))
print("Training data MSE = {0}".format(trainRegressionMetrics.meanSquaredError))
print("Training data RMSE = {0}".format(trainRegressionMetrics.rootMeanSquaredError))
print("Training data R-squared = {0}".format(trainRegressionMetrics.r2))
print("Training data MAE = {0}".format(trainRegressionMetrics.meanAbsoluteError))
print("Training data Explained variance = {0}".format(trainRegressionMetrics.explainedVariance))
print("Validation data MSE = {0}".format(validRegressionMetrics.meanSquaredError))
print("Validation data RMSE = {0}".format(validRegressionMetrics.rootMeanSquaredError))
print("Validation data R-squared = {0}".format(validRegressionMetrics.r2))
print("Validation data MAE = {0}".format(validRegressionMetrics.meanAbsoluteError))
print("Validation data Explained variance = {0}".format(validRegressionMetrics.explainedVariance))
print("Feature importances:\n{0}\n".format("\n".join(map(lambda z: "{0} = {1}".format(str(z[0]),str(z[1])), zip(featureCols, featureImportances)))))
.select("id", "prediction")\
.withColumnRenamed("prediction", "loss")\
.option("header", "true")\

Finally you can upload your result file to Kaggle if you want submit your results.



Feature Engineering with PySpark in Azure Databricks

Azure Databricks is a very cool easy to use platform for both analytics engineers and machine learning developers. I would like to use this post to summarize basic APIs and tricks in feature engineering with Azure Databricks. Previously I was using Jupyter notebook or PyCharm to develop or practice some machine learning cases. Compared to Jupyter Notebook, Azure Databricks provides similar Notebook, which adds some additional features. In the future I will describe how to use Azure Databricks for production.

Feature engineering is the preprocessing phase in machine learning, also needs huge effort from developers to get data ready for modeling and training. Here I list some basic feature engineering scenarios with PySpark in Azure Databricks.

If you want to run code snippet below in normal Jupyter Notebook, you need add Spark initialization code as below. (Not necessary in Azure Databricks, as it is already for use.)

spark = SparkSession\

Continuous Data

from __future__ import print_function
from pyspark.sql import SparkSession
from import Binarizer

continuousDataFrame = spark.createDataFrame([
(0, 1.1),
(1, 8.5),
(2, 5.2)
], ["id", "feature"])

binarizer = Binarizer(threshold=5.1, inputCol="feature", outputCol="binarized_feature")

binarizedDataFrame = binarizer.transform(continuousDataFrame)

print("Binarizer output with Threshold = %f" % binarizer.getThreshold())
from import Bucketizer

splits = [-float("inf"), -0.5, 0.0, 0.5, float("inf")]
data = [(-999.9,), (-0.5,), (-0.3,), (0.0,), (0.2,), (999.9,)]
dataFrame = spark.createDataFrame(data, ["features"])

bucketizer = Bucketizer(splits=splits, inputCol="features", outputCol="bucketedFeatures")

# Assign buckets per the splits boundary
bucketedData = bucketizer.transform(dataFrame)

print("Bucketizer output with %d buckets" % (len(bucketizer.getSplits())-1))
from import QuantileDiscretizer

data = [(0, 18.0), (1, 19.0), (2, 8.0), (3, 5.0), (4, 2.2), (5, 9.2), (6, 14.4)]
df = spark.createDataFrame(data, ["id", "hour"])
df = df.repartition(1)

# Divide into 3 buckets as quantile distribution
discretizer = QuantileDiscretizer(numBuckets=3, inputCol="hour", outputCol="result")
discretizerModel =
result = discretizerModel.transform(df)
from import MaxAbsScaler
from import Vectors

dataFrame = spark.createDataFrame([
(0, Vectors.dense([1.0, 0.1, -8.0]),),
(1, Vectors.dense([2.0, 1.0, -4.0]),),
(2, Vectors.dense([4.0, 10.0, 8.0]),)
], ["id", "features"])

scaler = MaxAbsScaler(inputCol="features", outputCol="scaledFeatures")

# Calculate Max value model
scalerModel =

# Transform with scale model, so that values are scaled between [-1.0, 1.0]
scaledData = scalerModel.transform(dataFrame)"features", "scaledFeatures").show()
#Standard scaler
from import StandardScaler

dataFrame ="libsvm").load("data/mllib/sample_libsvm_data.txt")
scaler = StandardScaler(inputCol="features", outputCol="scaledFeatures",
withStd=True, withMean=False)

# Calculate mean and variance
scalerModel =

# Standardize
scaledData = scalerModel.transform(dataFrame)
#Polynomial expansion
from import PolynomialExpansion

df = spark.createDataFrame([
(Vectors.dense([2.0, 1.0]),),
(Vectors.dense([0.0, 0.0]),),
(Vectors.dense([3.0, -1.0]),)
], ["features"])

polyExpansion = PolynomialExpansion(degree=3, inputCol="features", outputCol="polyFeatures")
polyDF = polyExpansion.transform(df)

Discrete Data

# StringIndexer and One-hot encoder
from import OneHotEncoder, StringIndexer

df = spark.createDataFrame([
(0, "a"),
(1, "b"),
(2, "c"),
(3, "a"),
(4, "a"),
(5, "c")
], ["id", "category"])

stringIndexer = StringIndexer(inputCol="category", outputCol="categoryIndex")
model =
indexed = model.transform(df)

encoder = OneHotEncoder(inputCol="categoryIndex", outputCol="categoryVec")
encoded = encoder.transform(indexed)


#Use Stop Words
from import StopWordsRemover

sentenceData = spark.createDataFrame([
(0, ["I", "saw", "the", "red", "balloon"]),
(1, ["Mary", "had", "a", "little", "lamb"])
], ["id", "raw"])

remover = StopWordsRemover(inputCol="raw", outputCol="filtered")
# Tokenizer
from import Tokenizer, RegexTokenizer
from pyspark.sql.functions import col, udf
from pyspark.sql.types import IntegerType

sentenceDataFrame = spark.createDataFrame([
(0, "Hi I heard about Spark"),
(1, "I wish Java could use case classes"),
(2, "Logistic,regression,models,are,neat")
], ["id", "sentence"])

tokenizer = Tokenizer(inputCol="sentence", outputCol="words")

regexTokenizer = RegexTokenizer(inputCol="sentence", outputCol="words", pattern="\\W")

countTokens = udf(lambda words: len(words), IntegerType())

tokenized = tokenizer.transform(sentenceDataFrame)"sentence", "words")\
.withColumn("tokens", countTokens(col("words"))).show(truncate=False)

regexTokenized = regexTokenizer.transform(sentenceDataFrame)"sentence", "words") \
.withColumn("tokens", countTokens(col("words"))).show(truncate=False)
# Count Vectorizer
from import CountVectorizer

df = spark.createDataFrame([
(0, "a b c".split(" ")),
(1, "a b b c a".split(" "))
], ["id", "words"])

cv = CountVectorizer(inputCol="words", outputCol="features", vocabSize=3, minDF=2.0)
model =
result = model.transform(df)
from import HashingTF, IDF, Tokenizer

sentenceData = spark.createDataFrame([
(0.0, "Hi I heard about Spark"),
(0.0, "I wish Java could use case classes"),
(1.0, "Logistic regression models are neat")
], ["label", "sentence"])

tokenizer = Tokenizer(inputCol="sentence", outputCol="words")
wordsData = tokenizer.transform(sentenceData)

hashingTF = HashingTF(inputCol="words", outputCol="rawFeatures", numFeatures=20)
featurizedData = hashingTF.transform(wordsData)

idf = IDF(inputCol="rawFeatures", outputCol="features")
idfModel =
rescaledData = idfModel.transform(featurizedData)"label", "features").show()
# NGram
from import NGram

#Hanmeimei loves LiLei
#LiLei loves Hanmeimei

wordDataFrame = spark.createDataFrame([
(0, ["Hi", "I", "heard", "about", "Spark"]),
(1, ["I", "wish", "Java", "could", "use", "case", "classes"]),
(2, ["Logistic", "regression", "models", "are", "neat"])
], ["id", "words"])

ngram = NGram(n=2, inputCol="words", outputCol="ngrams")

ngramDataFrame = ngram.transform(wordDataFrame)"ngrams").show(truncate=False)


# SQL Transformer
from import SQLTransformer
df = spark.createDataFrame([
(0, 1.0, 3.0),
(2, 2.0, 5.0)
], ["id", "v1", "v2"])

sqlTrans = SQLTransformer(
statement="SELECT *, (v1 + v2) AS v3, (v1 * v2) AS v4 FROM __THIS__")
# R formula transform
from import RFormula
dataset = spark.createDataFrame(
[(7, "US", 18, 1.0),
(8, "CA", 12, 0.0),
(9, "NZ", 15, 0.0)],
["id", "country", "hour", "clicked"])

formula = RFormula(
formula="clicked ~ country + hour",

output ="features", "label").show()

Note that: There are many feature engineering APIs are not mentioned here. Please see more info from latest package.


Batch Data Ingest with Azure Databricks

In the previous blog we have introduced basic steps of data ingest of streaming data with Azure Databricks. Now we are going to describe how to do batch ingest with Azure Data Bricks. Most of the steps are similar, just the spark DataFrame API are different.

Data Source

  "your blob account key",
  "your blob account key value here")
val df ="wasbs://yout container@your blob")
val availabilitySchema = df.schema

Please note that we need to df.schema to get availability log’s schema, similar for other log formar, e.g. page views, event. It is very hard to program the schema manually.

Data Transform

Use to get the Data Frame from input data. Please pay attention to input file format. inputPath/2018-05-30/09 should be matched like (inputPath + “/*/*/”)

val availabilityDF =
.option("maxFilesTrigger", 1).json(inputPath + "/*/*/")

Use Spark SQL to do the transform, much easier than transform with Data Frame API.

val sqlDF = spark
.sql("SELECT as ID, availability[0].testTimestamp as timestamp, availability[0].testName as name, availability[0].runLocation as location, availability[0].message as message, availability[0].durationMetric.value as duration FROM logs")

Data Sink

Here we use Azure SQL Server as the persistence of transformed data.

First define the database configuration

val jdbcUsername = "Your SQL Server username"
val jdbcPassword = "Your SQL Server password"
val jdbcHostname = "Your SQL Server" //typically, this is in the form or
val jdbcPort = 1433
val jdbcDatabase ="Availability"
val driver = ""
val jdbcUrl = s"jdbc:sqlserver://${jdbcHostname}:${jdbcPort};database=${jdbcDatabase};user=${jdbcUsername};password=${jdbcPassword}"

Setup connection properties for JDBC

import java.util.Properties

val connectionProperties = new Properties()
connectionProperties.put("user", s"${jdbcUsername}")
connectionProperties.put("password", s"${jdbcPassword}")
val driverClass = ""
connectionProperties.setProperty("Driver", driverClass)

Finally use write.jdbc api to persistence transformed data into Azure SQL Server, quite cool.

sqlDF.write.jdbc(jdbcUrl, "aiavailability3", connectionProperties)

Create bitcoin address

Bitcoin and BlockChain are too popular to ignored by technical people.  Hence I also spend some hours to read some books, and do some coding. This blog explains basic steps to create bitcoin address.

Step 1: Generate private key (256 bit, 64 bytes)

import binascii as ba

import ecdsa

import hashlib

import random

def create_priv_key():

   digits = ['%x' % random.randrange(16) for _ in range(64)]

priv_key = create_priv_key()

print('1. Private key:%s' % priv_key)

Step 2: Generate public key

def build_pub_key(priv_key):

   buff = bytes.fromhex(priv_key)

   sk = ecdsa.SigningKey.from_string(buff,


   pub_key = b'\x04' + sk.verifying_key.to_string()

   return ba.b2a_hex(pub_key).decode()
pub_key = build_pub_key(priv_key)

print('2. Public key:%s' % pub_key)

Step 3: Calculate sha-256

def sha256(key):

    sha = hashlib.sha256(bytes.fromhex(key))

    return ba.hexlify(sha.digest()).decode()
r_3 = sha256(pub_key)

print('3. Public key sha-256:%s' % r_3)

Step 4: Calculate ripedm160

def ripemd160(key):

    rm ='ripemd160')


    return ba.hexlify(rm.digest()).decode()
r_4 = ripemd160(r_3)

print('4. Calculate Step 3 ripemd-160:%s' % r_4)

Step 5: Prefix address with protocol version

r_5 = '00' + r_4

print('5. Prefix Bitcoin protocol version :%s' % r_5)

Step 6: Calculate sha-256 twice

r_6 = sha256(sha256(r_5))

print('6. Calculate Step 5 with sha-256 twice:%s' % r_6)

Step 7: Validation

r_7 = r_5 + r_6[:8]

print('7. Append output last step into Step 5 for validation:%s' % r_7)

Step 8: Encode with base58

addr = base58.b58encode(bytes.fromhex(r_7))
print(‘8. Wallet address: %s’ % addr)


Recently I am working on platform analytics solution for, the open industry data platform from DNV GL. Data Ingest is one essential part for the whole data analytics end to end flow.

Application Insights(AI) logs are one of the key data source, hence we need build streaming analytics solution for such real time data source. Here we take availability logs data as example, you can also use other  AI data sources (PageViews….).

Export the data

There is a nice document for enabling the continues export from Application Insights.

Create a storage account

When you are creating it, I suggest to use “Standard-LRS“ with Cold tier. This is simply because we use storage account as a temporary storage place. (We will move into VERACITY container for storage in production)

Configure the export settings

In this case, we are exporting one type:

  • Availability (One way for in production monitoring data)

Once it is done and the telemetry data started exporting, you should be able to see folders in the blob container in storage account. One folder is mapping to one data type in above export setting.

Take a closer look at the folder content. Below screenshot shows the PageViews data between 12am to 1pm on 28th of Sept.

Please note that:

  • The date and time are UTC and are when the telemetry was deposited in the store – not the time it was generated.
  • The folders under Availability are stored in yyyy-mm-dd/hh/ structure.
  • Each blob is a text file that contains multiple ‘\n’-separated rows.
  • It contains the telemetry processed over a time period of roughly half a minute. It means that each folder will have about 100 to 120 files.
  • Each row represents a telemetry data point such as a request or page view.
  • Each row is an JSON document. The detailed data structure can be found at here.

With the nice Azure Storage Account Explorer, it is pretty easy to check the content of the blob file.

Please note that Application Insights also implemented the same logic as IP Anonymizationin Google Analytics. For all IP address that is collected by Application Insights, last octagon is anonymized to 0 (you can see the highlighted in above screenshot).

Read the data from blob

They are many ways to transfer the data out from storage account. Here we are going to use Azure Databricks.

Add one Azure Databricks service following here.

Add one cluster, choose the basic one following here.

Add one notebook:  named for example: Streaming Data Ingest

In order Azure Databricks can read data from blob storage, there are two ways: Databricks directly read blob storage through HDFS API; Or mount blob storage container into Databricks file system. See more information about how to access Blob Storage as here. We choose the second one.

Mount the storage folder into Databricks file system(DBFS), to mount a Blob storage container or a folder inside a container, use the following command:

  source = "wasbs://<your-container-name>@<your-storage-account-name><your-directory-name>",
  mountPoint = "/mnt/export",
  extraConfigs = Map("<conf-key>" -> "<conf-value>"))

Then you can check the folder data with command:

%fs ls /mnt

Initialize SparkSession:

import org.apache.spark.sql.functions._
import org.apache.spark.sql.SparkSession

val spark = SparkSession

import spark.implicits._

In Spark 2.0+, we prefer use Structured Streaming(DataFrame /DataSet API) in, rather than Spark Core API, but when we see the Availability log data, it is XML like format, with several hierarchy. This is not easy to programming define the Structure type. The tricks here is that we first read one blob file to get Schema, then use this Schema when we create the structured streaming.

"<your key>")
val df ="wasbs://<your container>@<your-storage-account-name>")
val availabilitySchema = df.schema

Using structured streaming here to access the files in blob folder, we are going to emulate a stream from them by reading one file at a time, in the chronological order they were created.

//Use file sink to store output data into blob storage again.
val inputPath = "/mnt/export"
val outputPath = "/mnt/output"
val streamingInput = spark.readStream.schema(availabilitySchema).option("maxFilesTrigger", 1).json(inputPath + "/*/*/")

Transform the data

Note that we do not need all availability data fields for analytics. Here we only pickup the fields that you need, as below (Needed fields are highlighted)

We use SQL to do the transform, then convert the DataFrame into DataSet, suggested if you need to do more analytics afterwards. In Structured Streaming, query is a handle to the streaming query that is running in the background.

val sqlDF = spark.sql("SELECT as ID, availability[0].testTimestamp as timestamp, availability[0].testName as name, availability[0].runLocation as location, availability[0].message as message, availability[0].durationMetric.value as duration FROM logs")
val sqlDS =[(String, String, String, String, String, Double)]
val query2 = 
.option("path", outputPath)
.option("checkpointLocation", outputPath + "/checkpoint/")

Check the output folder in blob, to make sure that files are handled one by one continously (in micro-batch).

Note that: By the time writing this article, Structured Streaming does not support JDBC sink natively. Hence we need to extend the JDBCSink by ourselves.