watson-certification

Practical examples for Watson Certification

I created this project to study before the Watson certification exam. I was following this guide released by IBM which details the exam objectives. Note that these answers are my own and have not been validated by IBM.

Due to a lack of time, I only coded examples with Watson Services I've never used before. I already passed my certification so I won't add anything more.

Check the project of Jeronimo De Leon for a more complete Watson certification study guide.

Installation

pip install -r requirements.txt

For seaborn to work, you need to install tkinter

sudo apt-get install python3-tk

Examples

To run the examples, you need to change the credentials in config.py with your own credentials.

Personality Insights

Run the example to analyze Trump personality with the following command:

python section3/personality/personality_insights_example.py

Trump Personality Analysis

The code uses this Donald Trump's speech in text format as input.

We can obtain the results in a CSV format that shows a percentage for each personality trait:

big5_agreeableness	facet_altruism	facet_cooperation	facet_modesty	facet_morality	facet_sympathy	facet_trust	big5_conscientiousness	facet_achievement_striving	facet_cautiousness	facet_dutifulness	facet_orderliness	facet_self_discipline	facet_self_efficacy	big5_extraversion	facet_activity_level	facet_assertiveness	facet_cheerfulness	facet_excitement_seeking	facet_friendliness	facet_gregariousness	big5_neuroticism	facet_anger	facet_anxiety	facet_depression	facet_immoderation	facet_self_consciousness	facet_vulnerability	big5_openness	facet_adventurousness	facet_artistic_interests	facet_emotionality	facet_imagination	facet_intellect	facet_liberalism	need_liberty	need_ideal	need_love	need_practicality	need_self_expression	need_stability	need_structure	need_challenge	need_closeness	need_curiosity	need_excitement	need_harmony	value_conservation	value_hedonism	value_openness_to_change	value_self_enhancement	value_self_transcendence	behavior_sunday	behavior_monday	behavior_tuesday	behavior_wednesday	behavior_thursday	behavior_friday	behavior_saturday	behavior_0000	behavior_0100	behavior_0200	behavior_0300	behavior_0400	behavior_0500	behavior_0600	behavior_0700	behavior_0800	behavior_0900	behavior_1000	behavior_1100	behavior_1200	behavior_1300	behavior_1400	behavior_1500	behavior_1600	behavior_1700	behavior_1800	behavior_1900	behavior_2000	behavior_2100	behavior_2200	behavior_2300	word_count	processed_language
0.4268979029309932	0.8588535687537024	0.7044867158268264	0.5960736895707534	0.9676444453908276	0.9965335147666688	0.3998040334289172	0.9797657117581102	0.9070143051529225	0.9869915635122488	0.873983697434251	0.5488031217574947	0.9124250982342645	0.8090432365930695	0.6096825298389489	0.8828473589475007	0.9857260415356135	0.11637254228155247	0.01331810918980908	0.5959055919648221	0.16756739770553347	0.9505346628850558	0.02812768408642924	0.02455406377266378	0.168454469161186	0.05865378787277559	0.05754389444307567	0.015989819272919148	0.975519694665306	0.8328642953776038	0.654203608334551	0.24646836689211954	0.05370347126940961	0.9915865625089948	0.7792134600310711	0.01694074812415025	0.04771121338963841	0.0030600042610637868	0.023556425868524244	0.05084809704946042	0.23156088075382475	0.6470820764350991	0.028222045621644876	0.15436495592219218	0.3881229129662421	0.03816743693922564	0.035899020822038996	0.11167207177920224	0.01957680225835251	0.37441584209106304	6.717299135493016E-4	0.11547777255812258	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	6832	en

With a bit of seaborn magic, we can visualize Trump's personality results a bit more clearly:

According to the Personality Insights documentation, people who score high on Sympathy are tender-hearted and compassionate. Seriously Watson, Trump's highest personality trait is SYMPATHY? ;) But hey, I do agree he clearly doesn't need love.

Tone Analyzer

Run the example to analyze the tone of a text with the following command:

python section3/tone/tone_analyzer_example.py

Angry Sarcastic Text Analysis

The example analyzes this customer complaint.

The Tone Analyzer will return a percentage value for each tone analyzed per sentence:

The text is divided into the following sentences:

Dear Birmingham Airport Authority

Would it be possible to install a louder, more annoying warning siren for the baggage carousels?

The Martian ray-gun sound that you have installed at present is almost, but not quite, enough to induce insanity in arriving passengers as they await their luggage.

When it fails to stop sounding, it comes very close.

Such as last night, when it went off for about 15 minutes straight (all the while the ground crew failed to push the "deliver bags" button to operate the conveyor).

We obtain the following results:

Sentence	Emotion Tone:Anger	Emotion Tone:Disgust	Emotion Tone:Fear	Emotion Tone:Joy	Emotion Tone:Sadness	Language Tone:Analytical	Language Tone:Confident	Language Tone:Tentative	Social Tone:Agreeableness	Social Tone:Conscientiousness	Social Tone:Emotional Range	Social Tone:Extraversion	Social Tone:Openness
0	0.174853	0.045767	0.203073	0.363684	0.288548	0.0	0.0	0.0	0.601976	0.274405	0.287173	0.549738	0.192534
1	0.413234	0.332002	0.11745	0.09863	0.157501	0.029341	0.0	0.615352	0.092194	0.01064	0.289132	0.227488	0.732911
2	0.103462	0.261834	0.461912	0.015445	0.308681	0.403089	0.0	0.5538	0.408464	0.509407	0.664799	0.305697	0.814446
3	0.240445	0.10141	0.188986	0.049759	0.529359	0.0	0.849827	0.0	0.620178	0.041014	0.039034	0.312732	0.112508
4	0.24863	0.144169	0.370541	0.075804	0.290018	0.85019	0.204269	0.0	0.002668	0.907311	0.915778	0.457429	0.078753

Well he's clearly open to change the Martian Ray-Gun alarm so I'm guessing that openness is right. It seems the results are a bit like the horoscope, you can pretty much find a reason why they fit for whatever gets predicted.

Visual Recognition

Classify images with the default ibm classifier:

python section3/vision/predict_ibm_classifier.py 

Train a custom fruit classifier.

python section3/vision/train_fruit_classifier.py 

Use your custom fruit classifier to classify images. You must add the classifier_id in the config.py file.

python section3/vision/predict_custom_classifier.py 

Is Chuck Norris a Fruit?

Spoiler: he's not.

The classifiers are tested on the following images:

IBM Pre-Trained classifier results

image	classes	score
FruitMan	vegetation, food	0.73, 0.57
ChuckNorris	person	0.83
GrumpyCat	animal, mammal, cat	1.00, 1.00, 0.98
UnicornMan	person	0.99

Custom Fruit Classifier results

image	class	score
FruitMan	fruit	0.53
ChuckNorris	not a fruit
GrumpyCat	not a fruit
UnicornMan	not a fruit

Sections

• Section 1 - Fundamentals of Cognitive Computing
• Section 3 - Fundamentals of IBM Watson Developer Cloud

Similar Projects

Check the project of Jeronimo De Leon for a more complete Watson certification study guide.