GCDEC/Deploying Tensorflow/Notes - Revision history

Admin at 10:21, 7 January 2018

2018-01-07T10:21:44Z

← Older revision		Revision as of 10:21, 7 January 2018
Line 907:		Line 907:
	[[Category:Data Engineering]]		[[Category:Data Engineering]]
	[[Category:Google Cloud]]		[[Category:Google Cloud]]
			[[Category:Tensorflow]]
	[[Category:ML]]		[[Category:ML]]
	[[Category:NN]]		[[Category:NN]]

Admin: /* Module 4: Feature Engineering */

2017-10-21T23:32:49Z

Module 4: Feature Engineering

Show changes

Admin: /* Scaling with Cloud Machine Learning Laboratory */

2017-10-21T23:20:34Z

Scaling with Cloud Machine Learning Laboratory

Show changes

Admin: /* Turning Raw Data into Numeric Features */

2017-10-18T13:23:00Z

Turning Raw Data into Numeric Features

@@ Line 483: / Line 483: @@
 (This data comes from a web app, goes into a data warehouse, and is pulled out as JSON data)
-Now turn this into a feature column:
+====Creating a feature column====
 * Some fields can be used directly (e.g., customer rating, price, and wait time)
 * Others (e.g., transactionId) should be ignored (don't have more than 5 examples)
@@ Line 497: / Line 499: @@
+]
 </pre>
 Preprocessing the data creates a new "vocabulary" of keys - and it needs to be available for BOTH training AND prediction steps (e.g., prediction sends employeeId 75534, and model needs to know how to convert that to a one hot encoding)
@@ Line 517: / Line 523: @@
 <pre>
-layers.sparse_column_hashbucket('employeeId', 500)
+layers.sparse_column_hashbucket('employeeId', 500) # Hash the employee ID, and break it into 500 buckets
 </pre>
 =References=

Admin: /* Enough Examples */

2017-10-18T13:06:31Z

Enough Examples

← Older revision		Revision as of 13:06, 18 October 2017
Line 439:		Line 439:

	Rule of thumb - need AT LEAST five examples for a category to be usable in an example		Rule of thumb - need AT LEAST five examples for a category to be usable in an example

			Quiz: which is difficult to have enough examples of?

			Predicting total number of customers who will use a coupon:
			* Percent discount of coupon - DEPENDS (find five examples each, or throw it out; if you have continuous numbers, bin them into discrete groups)
			* Date that promotional offer starts - DEPENDS (bin them up again, e.g., promotional offers starting in January or in Q1)
			* Number of customers who opened advertising emails - TRUE (should have a number of different emails, and know how many customers opened each)

			Predict whether CC transaction is fraudulent:
			* Whether cardholder has purchased this item at this store - TRUE (should have this, unless it is too specific, e.g., bought diapers between 8 and 9 pm)
			* Distance between cardholder and store - DEPENDS (again, should bin these up; may not have 5 examples of cardholders who bought something from a store more than 100 miles from their house)
			* Category of item being purchased - TRUE
			* Online or in-person purchase - TRUE

			How to check? Plot histograms of data.

			===Turning Raw Data into Numeric Features===

			Example: running ice cream store, want to predict the rating a customer will give based on how long they've been waiting and what they bought

			Raw data to TensorFlow feature column.

			Raw data:

			<pre>
			{
			"transactionId" : 42,
			"name" : "Ice Cream",
			"price" : 2.50,
			"tags" : ["cold", "dessert"],
			"servedBy" : {
			"employeeId" : 45042,
			"waitTime" : 1.4,
			"customerRating" : 4
			},
			"storeLocation" : {
			"latitude" : 35.3,
			"longitude" : -98.7
			}
			}
			</pre>

			(This data comes from a web app, goes into a data warehouse, and is pulled out as JSON data)

			Now turn this into a feature column:
			* Some fields can be used directly (e.g., customer rating, price, and wait time)
			* Others (e.g., transactionId) should be ignored (don't have more than 5 examples)
			* Some (e.g., employeeId) should be transformed - no meaningful magnitude... use one hot encoding
			*

			<pre>
			INPUT_COLUMNS = [
			...
			layers.real_valued_column('price'),
			layers.real_valued_column('waitTime'),
			...
			]
			</pre>

			Preprocessing the data creates a new "vocabulary" of keys - and it needs to be available for BOTH training AND prediction steps (e.g., prediction sends employeeId 75534, and model needs to know how to convert that to a one hot encoding)

			Three example scenarios:

			First scenario: you already know the keys beforehand (e.g., employeeId and one hot encoding):

			<pre>
			layers.sparse_column_with_keys('employeeId', keys=['12345', '48506', '28488', '23456'])
			</pre>

			Second scenario: your data is already indexed 0 to N, but does not have a meaningful magnitude (e.g., hour of the day):

			<pre>
			layers.sparse_column_with_integerized_feature('employeeId', 5)
			</pre>

			Third scenario: you don't have a vocabulary of all possible values:

			<pre>
			layers.sparse_column_hashbucket('employeeId', 500)
			</pre>

	=References=		=References=

Admin: /* Numerical */

2017-10-18T12:37:09Z

Numerical

← Older revision		Revision as of 12:37, 18 October 2017
Line 433:		Line 433:

	If you do an arbitrary item categorization, you lose meaning and qualities of words (e.g., male/female, soft/hard, positive/negative, etc.). So, word2vec greatly improves the ability of word inputs to help improve the prediction.		If you do an arbitrary item categorization, you lose meaning and qualities of words (e.g., male/female, soft/hard, positive/negative, etc.). So, word2vec greatly improves the ability of word inputs to help improve the prediction.

			====Enough Examples====

			Each feature needs enough examples to be ''understandable in context''

			Rule of thumb - need AT LEAST five examples for a category to be usable in an example

	=References=		=References=

Admin: /* Module 4: Feature Engineering */

2017-10-18T12:36:27Z

Module 4: Feature Engineering

@@ Line 356: / Line 356: @@
 ==Module 4: Feature Engineering==
 Feature engineering and pre-processing using Cloud ML - ways of making our data set better
@@ Line 363: / Line 365: @@
 Still have the original TensorFlow model - improve it using feature engineering and hyperparameter tuning
-Good freatures:
+Good features:
 * Related to objective - reduce arbitrary data
 * Known a priori
@@ Line 369: / Line 371: @@
 * Enough examples
 * Bring human insight (domain expertise) into probem
 Related to objective:
@@ Line 386: / Line 390: @@
 * Category of item being purchased (TRUE)
 * Expiry date of credit card (FALSE)
 =References=

Admin: /* References */

2017-10-18T09:35:21Z

References

@@ Line 354: / Line 354: @@
 * Deploy the model to the cloud
 * Call the model to make predictions
 =References=

Admin: /* Scaling with Cloud Machine Learning Laboratory */

2017-10-18T09:17:16Z

Scaling with Cloud Machine Learning Laboratory

← Older revision		Revision as of 09:17, 18 October 2017
Line 343:		Line 343:

	===Scaling with Cloud Machine Learning Laboratory===		===Scaling with Cloud Machine Learning Laboratory===

			Lab link: https://codelabs.developers.google.com/codelabs/dataeng-machine-learning/index.html?index=#9

	Use a single-region bucket for machine learning training inputs and outputs - enables consistency (fast reading/writing from multiple threads)		Use a single-region bucket for machine learning training inputs and outputs - enables consistency (fast reading/writing from multiple threads)

Admin: /* Pick Up Here */

2017-10-18T07:16:37Z

Pick Up Here

← Older revision		Revision as of 07:16, 18 October 2017
Line 352:		Line 352:
	* Deploy the model to the cloud		* Deploy the model to the cloud
	* Call the model to make predictions		* Call the model to make predictions

	~~=Pick Up Here=~~



	~~==Module 4: Feature Engineering==~~

	=References=		=References=