OK, so now’s the time to fess up. Yes, it’s shocking, but true. I know this is going to rock some of your foundations to the core, will rob the rug of cosy childhood memory from under your feet and stop you mid-sentence, but in the interests of upholding the mores of a pre-post-truth era, it still has to be said, and I can now reveal with 95% confidence, it appears that:
…..Wally is actually a woman….
Here is the hard-core evidence witnessed when watching the videoed recording of Demis Hassabis of Deep Mind, talking recently to a seated audience of entrepreneurs at the Google Campus, London.
I have also been doing my damnedest to get a (highly scaleable, transactional and disruptive) tech startup started….the idea itself has been nominated for an award, so it’s evidently not shit, but you try touting it around the startup community as a woman of a certain age, with a certain lack of funds and well, Wally feels pretty elusive!
This is so crazy, as we already know that:
Studies have shown that diverse teams create better products and run more successful companies. We also know that women are highly efficient with capital, running companies on 2/3 the funds of their male counterparts, yet receiving roughly 3 percent of venture capital. Yet women are still underrepresented in startup communities.
What? Why? Where is the logic there? So, if the guys are the logicians, where does that leave us? Oh yeah, we must be the magicians! Well, I have to say, where Women in Kaggle is concerned, yes we are!
The datadive – amazing! Women in Kaggle were a triumph. I circulated an idea, found a suitable dataset, asked for 5 volunteer mentors (one goto resource per group) equipped each with a focused and probing business question, and wow – did they step up and deliver. Unpaid, over worked and undervalued (can you believe? – certainly not by us!). Outside their day jobs, and for the good of their community, these ladies arrived, armed with a meticulous plan of attack, code, cookies and cake…and that was damned good cake…
On the subject of Wally, I have a sneaking suspicion that Randal Olson’s: “machine learning system that works its way through the first seven Where’s Wally books and quickly spots the character.” might just struggle to spot her given the training data, but hey, what’s new in the world of beardless tech?!
So, a quick resume of our datadive efforts of which I am, and WIK ladies should all be, so proud! I knew a lot of WIK members were being approached by companies seeking attribution modelling and general customer segmentation skills, so in response to popular demand we held our very own WIK datadive to dig into just those areas:
Working with a retail dataset, we set out to address the following business problems:
1. Pinpointing key trends in your business to focus your marketing investment
2. Identifying your best customers so you can maximise engagement with the most valuable segment
3. Using profiling to understand which new customers are likely to become loyal repeat purchasers
4. Leveraging insights to action via a multichannel communications strategy
We wanted to share the collective WIK experiences of couple of weeks back, at the Bloomberg sponsored DatakindUK datadive, at our next meetup. We hoped to recreate the kind of approach and problem solving that we saw at the datadive, which emerged from exploring the real world questions of the contributing organisations – perfect fodder for interviews questions and for growing current Data Science skills in a multitude of ways! I put together some slides about exploring customer segmentation type problem/questions to present and distribute.
I chose a real-world dataset whose dissemination criteria was restricted and therefore only distributed within the group under its educational auspices and as approved by the data owners. The dataset contains household level transactions over two years from a group of 2,500 households who are frequent shoppers at a retailer. It contains all of each household’s purchases, not just those from a limited number of categories. For certain households, demographic information as well as direct marketing contact history are included.
And provided some general customer profiling questions for everyone to ponder in the meantime….(and boy, did they ponder!…)
1.1 Recognising customer groupings/profiles:
Based on demographics, are there recognisable profiles/ groupings of customer who buy from our store?
1.2 Changes in customer profiles over time:
Are there identifiable changes over time in the profiles/ groupings of customer buying from our store? If so, what changes can be seen?
1.3 Identifying client profiles against product type:
Do customer profiles differ depending on the type of products they buy?
We worked separately in 5 teams of 8, with one mentor and one, more specific data driven question per team. We reconvened over delicious pizza from H2O and sublime cookies and cake from Mia to share a cohesive picture, and then…. after the break, pieced it together to construct a 360 view of our customer base and gather insights into their behaviour.
It was incredible, 5 groups and 5 solutions, each building on former sessions, drawing together our collective knowledge, forging forward as a tight knit, supportive group, the mentors (yay for Suzanne who expertly steered an entire herd towards newbie nirvana!) ensuring that nobody got left behind…Unbelievable, we covered introductions, presentations, problem solving, socializing, eating, drinking, animated discussion, furious coding, more presentations and thoughtful, collective insights, morale boosting, networking, job searching opportunities and tips, interview reviews and tips, expert githib tuition and tips (thanks Tzhe’ela!) and FUN!!!!
As the datadive was a huge success, we would like to extend it to a weekend day in addition, and are looking either for paying companies to sponsor a hotbed of talent to address their business problems, or a not for profit who are looking for us to help them address recognized or unrecognized data problems within their organisations. You get your problems solved and we get to enhance our CVs! Please contact us with suggestions/ideas at firstname.lastname@example.org. We hope to spend a day building you a ‘shiny’ (literally) management tool with which to visually interrogate your data.
And so, this was our sixth meetup and I’m very proud of how our group has grown over these past six months! It was great to have an opportunity to thank everyone who’s helped out by awarding them their very own Kaggle mugs!
Wonderful mentors and speakers, from left to right: Shappy, Anuiska, Suzanne, Chiin and Georgie (and the much missed Vicky and Kim who will collect theirs very soon! And to thanks to Nicola from Thoughtworks for hosting our event!)
So onward and upwards ladies! And let’s not forget….
Oh, and before I close, I keep getting requests from guys to join Women in Kaggle, so I just wanted to explain that it is a group for women, not in the slightest because I wish to exclude men, but simply because I am doing my little bit to offer an opportunity for women who have been hugely under-represented, not only in the world of tech, but let’s just say in the world, full stop. I have a capacity of 40 seats for 40 bums and I would like these bums to belong to women, because in every other area of their life that day, until they arrive at that particular venue that night I’ll wager that they will have been under-represented, and I really don’t think that in a city of nearly 9 million, my setting aside 40 seats is really going to disrupt the 4 and a half million of whom are male.
Whereas, in reaching out to one vulnerable woman who is attempting to dip her toe in the river of independence, boosting the confidence and skills of one self-effacing lady who has never considered asking for a pay rise, let alone equal pay, might, just might actually equip her to speak out and might, just might change her world…..
please understand that guys!
Keep Kaggling till the next time,