How Do I Get a Job Doing Data Science If I Don't Have Much Experience?

I regularly get calls from MBA grad students with a similar situation: "I took a few data science classes, love it, and want to get a job doing data science. However, my resume doesn't show much, if any, related experience. How do I get a job doing data science?"

It's a great question, and it makes sense why they'd call me. I was in that exact position when I graduated. The only difference was at the time the term "data science" didn't exist and tools were still in their infancy (Google had just bought Youtube, and we were all still wondering if this "video upload" site thing was viable).

In 2010 I wrote about a general post about finding a job, which is still relevant, and here's my specific answer to aspiring data scientists.

1. Define Your Talents

The Talents

Data science, like Marketing and HR, is a broad term with many jobs and skills within. So when you tell me, a person looking to hire data scientists, you want to be a data scientist, I'm still not sure what you mean. To which I usually follow up with the profound question of, "What does that mean to you?" I typically get a bland response simply because they're still discovering what it all means themselves. Here's some guidance:

First, decide if you're going to be a specialist or semi-generalist. A specialist is easy: focus on one thing and be the best in the world at it. Get amazing at producing accurate forecasts. Become THE data cleaning person. A warning: you'll be competing against PhDs in this space.

My recommendation is to go down the semi-generalist path. Find an interesting combination of skills/interests that's helpful. You're good at project management AND forecasting? That's interesting. You crunch numbers AND can give a presentation to executives? Now you have my attention.

HBR wrote a fantastic article called Data Science and the Art of Persuasion, and I recommend reading it at least once. It breaks data science into six talents. In brief, here they are:

1. Project management: do you know agile/scrum project management? Can you be an effective scrum master and get product owners to define done and prioritize?
2. Data wrangling: can you find, clean, and structure data? Are you good at automating processes?
3. Data analysis: can you find meaning in data and apply it to a business question?
4. Subject expertise: what business model or industry are you interested in?
5. Design: are you good at data visualization? There's the technical side, but also the art side of it. 
6. Storytelling: Can you write a narrative around the data and analysis? In my experience, this is the weakest talent on a data science team.

The People

If you read that list and still aren't sure, go talk to people. Actually, no matter what, go talk to people. Informational interviews are amazing. It's the lowest risk way to learn about something (which, if you messaged me to chat, and I asked to you read this first, and you are right now... good! keep going!).

Find other data scientists. Find people who work with data scientists. Find people who work in data-intensive industries.

I go into more depth in my original post, but here's the high level: come up with a set of questions to ask everyone. You may not ask all of them, but having a pre-set list will give you the confidence to reach out because you're prepared. Talk to at least 12 people, more if you can swing it. I talked to 2-3 per week while in school.

The Companies

My observation is there are three types of companies. They all have pros and cons. After talking with people and evaluating your talents, one of these types of companies may attract you.

Large companies, especially in data-driven industries, typically have a dedicated data science team (though the name might be different). Or, at least a specific role dedicated to data analysis. All fortune 500, if not 5,000 have this setup. These types of jobs tend to have narrow scopes. For example, my job at HP is to forecast how much money HP will make selling ink. Very specific. I use all 6 talents to do it, but my scope is narrow.

Small to medium size companies will have a small marketing team, but no data scientists. In my opinion, these are interesting positions. You don't need to be the best analyst in the world, but you can bring your skills and add a lot of value to the original position. For example, my internship was with a 500-person credit union. My job was to create a new checking account offering. I did competitive research, ran a survey... and did an analysis on their customer database. Nobody else on the team knew how to access this fantastic resource! Granted, I had to figure out how to get access and build my own tools (data wrangling), but I was OK with that. It was like I had superpowers relative to the rest of the team. The pay is typically less, but the variety is more, and the expectations are lower. This could be a perfect first step.

Boutique analysis consultants are super small, 1-5 people, companies that specialize 100% on data analytics. Like a typical consultant, they come into a company, learn about their problem, and do an analysis on behalf of the company. These are harder to get into, but you can learn a ton as an apprentice in a short amount of time. Be willing to do the grunt work for long hours.


You think you know what you want to do, or at least have a plan to figure it out? Great! Let's move on.

2. Cultivate Your Talents Into Skills

"The separation of talent and skill is one of the greatest misunderstood concepts for people who are trying to excel, who have dreams, who want to do things. Talent you have naturally. Skill is only developed by hours and hours and hours of beating on your craft." - Will Smith

The lack of experience on your resume is tough. Unfortunately, there's no shortcut. To get the experience, you actually need to practice data science. The good news is you don't need a job to "beat on your craft". There's almost an unlimited amount of data thanks to the internet. Pick a subject, find some data, and do an analysis.

FiveThirtyEight has its own dataset and often points to its original sources within an article. Sports data is interesting, so is political/polling data. There's census data, economic data, Zillow home sales data, stock market data, and so much more! There's data specific to an industry you're interested in. Collect your own data!

I tend to collect my own data and pick ones relevant to my life.

• I ran a half marathon, and they posted the results online. So I turned it into an infographic. I'm particularly fond of this one.
• I tracked my steps, weight, and food. Then I analyzed it and shared what I found.
• I created a map of real estate sales in my area. I thought this was particularly cool.
• I gave a transparent look at our investment property's performance. I still get calls about this one.
• I tracked my steps
• I tracked more steps
• I found out I could copy and paste my iTunes library into Excel. So I had fun with it.
• While visiting family in Port Townsend, I learned they get less rain than we do. So I found weather data and wrote a post about it.
• A random one about the US Government.
• A follow up to a podcast topic about PC vs. tablet shipments.
• Once I even did a couple of posts about my personal finances.
• I also did some tutorials, like a DIY loan calculator.

Can you guess my talents? I like the design and storytelling side. I'm much lighter on the technical data analysis and data wrangling parts.

If I were looking to get a job, I would do at least one per week. Beat on your craft to hone your skills and gain experience. We'll talk about what to do with all this effort next.

There is one more way to gain experience: [pro-bono] freelance work. There are a bunch of small businesses and organizations that would love help analyzing their data, but they don't know how to do it (like my credit union example), and often times they can't afford to hire someone to do it. This is an opportunity!

For example, I talk to contractors and landlords all the time who would LOVE to create an annual budget but don't really know how to go about it. Volunteer to take a look and help. I created a forecasting tool in R for my church to better predict giving trends. I worked with a local broker to create a "state of the real estate market" report he could share with clients. It didn't pay much, but it led to future projects. In fact, don't do it to get paid. Do it to help, and then add it to your resume.

Another interesting option for freelance work is to signup for Upwork. It's a website that companies go to for one-time projects and/or part-time work. Often times it's remote. I've hired 4 people from the site, and it's fantastic. Tip: I've noticed that data scientists call themselves a "financial analyst" on the site. If anything, it'll give you interviewing experience.

Got a plan in place to cultivate your talents? Let's show it off.

3. Showcase Your Skills

In the graphic design world, people have portfolios. In the software development world, people have gits. But they're the same thing: a way to showcase your work. As a data scientist, you want to create the same thing.

Think about it from an employer perspective: I'm about to take a considerable risk, financially and culturally, to bring someone into my organization. I want to reduce as much uncertainty as possible. That's why employers rely so heavily on recommendations and another reason why informational interviews are so important.

Reading accomplishments on a resume are helpful, but reduce my risk a lot and SHOW me what you did. That's why designers have portfolios. I don't care what school you went to, I want to see for myself you know how to draw.

So, create a WordPress site and find a theme that works for you. Or whatever tool you want - don't overthink this. Upload your projects to Google Drive and share the folder. It's not as pretty, but I'm not evaluating you on how pretty it is.

Then, as you do projects, add it to your website. Start with every. Single. Homework. Assignment. At this point in the game, you need volume, you can add a featured section later. But don't just post the homework, take one more step that'll blow everyone out of the water: create a short video where you walk through what you did. Wrote some R code? Use Zoom to share your screen and talk through it. Embed the video using Youtube (it turns out it was viable) along with the code. Or put the code in Github and link to it.

Did you have a final project with a presentation? Record yourself giving the presentation (do it again by yourself if the presentation came and went). Then record again showing the actual analysis. Show your work. For group projects point out the work you did.

I use video all the time, and it's fantastic. When I make an offer on a property, I don't just kick over the offer and hope the seller can figure out my thinking. I record a quick video on Zoom, upload it to Youtube and share it with the document. I get comments all the time about how professional it looks. You should do the same thing. I personally choose to keep the video on so they can see me and my expressions, but you don't have to. This is especially important if your talent is storytelling.

I made a sample presentation so you can see exactly what I'm talking about. Mine was done off the cuff. In reality, I'd script out what I wanted to say a lot more... and slow down my talking. Anyways, here it is:



You can also make tutorials. You can comment on other people's work. Flowing Data is a perfect example of these types of posts. So is Edward Tufte's Twitter account. If you do the commentary, go further than they do. Go into detail on what's good and bad.

Then put the link to your portfolio on your resume.

When following up after an informational interview, you might mention a project you did, share the link, and ask for their feedback on how to make it better. Then... follow their advice to improve it.

4. The Briefcase Technique

If you do this, you'll get interviews. Now it's time to knock it out of the park. Part of the interview preparation process is researching the company. Take it one step further and do an analysis for the company. The closer you can get to what you'll actually be working on, the better. If you interview for me, look for IDC data on printers. If you can't do that, look at HP's historical revenue.

Imagine being able to say, "I was looking at IDC shipment sales, and I found that the market is shrinking. Here, let me show you. (take a paper with a chart out of a briefcase, or portfolio, whatever).  I noticed X Y Z... And a question." Or, "It looks like HP is under-represented in the copier space. I recommend investing more there." Or, "why isn't HP in the copier space?"

I learned about this technique for freelancing from Ramit Sethi, but it works just as well when interviewing for a data science job (hint: it actually works for all positions).

Look, you're probably not going to teach me something or uncover some new insight. That's not the point! Your goal is to show me you care, to show off your skills, and to demonstrate you're going to take the extra step of a top performer.

Final Thoughts

I know what I'm suggesting is a lot of work, but the results will be worth it. Plus, it's a low-risk way for you to discover if you genuinely like data science without having to commit to a job to find out.

Good luck with your job search!


Photo by Carlos Muza on Unsplash

How I lost 18 Pounds in 37 Days

I'm training for a Tough Mudder next weekend. A Tough Mudder is a 10 mile run with 20 obstacles in your way. It's a combination of cardio, strength, fear factor, and group effort. I'm excited and nervous. My training goals are simple: run 10 miles non-stop, do 40 push-ups, and weigh 165lbs. That would put me in enough shape to enjoy the event. Here's a trailer for the event:


So, I started running semi-regularly in the middle of April. My basic training method is two short runs in the middle of the week (3-4 miles) and a progressively longer run on the weekend (add 1 mile each week). Here's how I did:

Despite starting to run regularly, my weight continued to go up! So being a data person, I started tracking my calories (via MyFitnessPal) in addition to my weight (via a Withings Scale) and my motion (via Apple Watch). Here are the results (all the time periods line up):

Some observations:

• I hit my goal of 165 lbs and running 10 miles non-stop! (I can also do 40 push-ups) I am very happy I won't be carrying those extra 18 pounds with me over giant walls.
• The day I started tracking my weight, my weight started falling. That's how my body's built. I hated it when playing football, but now it's good.
• Despite regularly increasing my miles, my peak calories burned stayed about the same. In other words, I "made up for it" the rest of the day by not moving as much.
• There are days when I don't eat a lot. For example, yesterday I skipped breakfast and lunch; not on purpose, it just happens sometimes. MyFitnessPal yells at my when that happens.
• My average weight loss has been 0.46 lbs per day, but can swing as much as 3 lbs each day! I'll be honest, this can be a little bit of an emotional roller coaster.

So? How did I lose 18 pounds in 37 days? I ran a bunch and ate less. But there's more behind it.

The Science

Science tells us that if you spend more calories than you consume, you'll lose weight. Now, how fast you lose weight and how much is muscle vs fat depends on the type of food you eat, how big the calorie deficit is, and the type of movement you do (strength vs cardio vs nothing).

As a general eating rule: eat/drink almost zero sugar, eat less carbs (breads & fruit), and watch your dairy. Protein and fats (ironically) are good for you because your body stores that energy differently. Whole foods are your friend. If you want to lose weight fast, good examples are the Slow Carb diet and Keto diet. I'm not following either of them strictly, but staying close to the principals.

As for exercise: try to incorporate movement into your day and strain yourself regularly to remind your body you need your muscles. If you can, ride your bike, stand at your desk, and walk places. You actually don't need to train for a marathon. According to FiveThirtyEight, the 5K, not the marathon, is the ideal race. The trick is to push yourself for speed; to exhaustion, and be sore when you're done. You can do this in as little as 4-7 minutes a day. Seriously. You also don't need a gym. For me, I find I need some sort of external motivation - like training for a Tough Mudder - to get me moving.

The Statistics

That's the science, which I know enough about to be dangerous. But statistics is my jam... and since I had the data... :)

I created a regression model to see which factor: Food, Motion or Runs, correlated the best. In other words, do I just need to watch what I eat? Can I make sure to hit a move goal by the end of the day? Or do I need some sort of intense workout? Or some combination?!

Here's what I found for correlations ( 1 = perfectly correlated. + means same direction. - means opposite direction):

• Food: +.61
• Motion: -.33
• Running: -.32

It appears you are what you eat. What you eat is twice as important as how much you move! You can visually see it in the very first chart.

I took the food one level deeper and looked at my macro-nutrients. Can I eat whatever I want as long as I watch the amount? Or, do I also need to pay attention to what I eat? Here's what it looks like:

Here's what I found for correlations:

• Carbs: +.54
• Fat: +.07
• Protein: +.002 

Well... Well... Well... It looks like science knows what it's doing. Carbs are bad and should be substituted for fat and protein when possible.

Key Findings

Want to lose weight? This should sound familiar:

Focus on your diet. Eat less in general, and really focus on eating less carbs. If you reduce your carbs, you'll naturally reduce the amount of sugar you eat. Use an online calculator or an app like MyFitnessPal to determine your daily target calories.

You should also exercise. You don't need to do extreme workouts, but focus on hitting a total movement goal (like 10,000 steps). Doing something of high intensity for a short amount of time is one way to get the overall goal, but isn't needed for weight loss (it is needed for building muscle).

It's nice to know my data supports the science. :)

Future Aspirations

My goal last year was to weigh 160 lbs and I didn't make it. I'm pretty close right now, so I'm going to go for it!

Finally, watching my weight is OK, but a measure I'm more interested in is percent body fat. If I can control this, my weight won't be an issue. Here's where I'm at right now (this, by the way, is why I have a Withings Scale):

According to the American Council on Exercise, I was average and just dipped into the "Fitness" category. If I get below 14%, I'll be an "Athlete".

My desire is to be below 15%. OK. My real desire is to get rid of a couple lovable handles. That way I'll be able to keep up with the kids as they get older. I don't do a lot of resistance training right now, and that'll need to change to help get this down. I have a big goal in mind of next year which should help me get there.

The Shocking Amount of Rain Corvallis Receives

So I'm in Washington visiting family. While downtown, a random person asks to pet Vinnie and says, "You know, it really doesn't rain that much in Port Townsend. I'm actually surprised it's as green as it is."

Wait. What?

Before this moment, my mental model was that it rained ALL THE TIME in Port Townsend. It goes something like this:

It never rains in San Diego.
Further North in Los Gatos it rains sometimes.
Further North in Corvallis is almost rains all the time.
Further North in Port Townsend it must therefore rain all the time.

Like any good data scientist, on Saturday morning before anyone else woke up, I visited Weather.com to see what the actual historical averages are. Here's what I found:

Well... well... well... It turns out the random lady, who likes doxies, was right. I'll resist the urge to connect petting dachshunds with weather soothsaying given my recent record.

The crazy part is that it rains LESS THAN HALF as much in Port Townsend compared to Corvallis.

Well, now that I had the tool open, I opened a spreadsheet and started looking up cities of interest. The chart below is total annual rainfall in inches.

• Los Gatos = Hometown
• San Diego = "doesn't rain much" and I know people who work there
• Boise = I know people who work there
• Bend = Where I tell my parents it's nice and they should move there
• Seattle = is Port Townsend special? (yes... to a degree)
• Newport = I needed to find one city with more rain than Corvallis
• CO Springs = family lives here

Ok. At this point, some of these look the same, but I think we can all agree that Boise is NOT the same as San Diego. Obviously a better picture would include temperatures, snow and/or sunny days, but that wasn't the claim I was investigating.

I did want to get a better feel for how the average monthly rainfall looked. So the following chart shows the average in inches with bars that span the min/max amounts of rain in a month.

So it might rain less overall in Port Townsend, but there are some months when Corvallis gets less (take that!).

And now for the series. This isn't for the faint of heart. I wanted to see the patterns. If you make it all the way to the end, you're in for a surprise in Colorado Springs.

Most follow the same pattern (more in winter, less in summer) and this shows how extreme it gets in each place. Surprisingly, there are months when it rains MORE in San Diego than Boise, but that's offset by lots of months with little, if any, rain.

I'm not 100% sure, but I think Colorado Springs is broken.

Pretty cool, right? I think I'll spend some time reminding myself why I live where I do...

Who Has Control of the US Government?

I was working on a fun project which involved taking a look at who controlled the House, Senate and Presidency each year for the past 50 years. Since I had the data sitting around, I decided to make a quick graphic of it. You can click on the picture to see a bigger version of it.

My goal was to convey who had control (the majority), but also show that often it's a very narrow majority each year. I'd love to hear your feedback on how to make it better. I'll make sure to share my final project too when it's done.

PC vs Tablet Shipments: Fun With Data

So, I got into a debate with my brother during one of our Furlo Bros podcasts over whether or not tablets (ie. the iPad) should be counted as a PC or not. What re-kindled the whole debate was Canalys' decision to include the iPad in Apple's PC shipment numbers. Which brought about the headline grabbing conclusion that Apple would surpass HP as the leading PC vender in 2012.

Matthew thinks the iPad should be counted. I do not.

With the latest Gartner PC shipment report, I decided to do some digging myself to see what the data shows. Gartner makes it fairly easy to look up each quarterly report and grab the data in semi-copy-and-paste friendly data tables. See below for all the links**. Here's what I found:

A couple notable points:

• HP is the current WW leader with 16% share of the market. Congrats team!
• HP also has some vicious Q4 swings in shipments. I bet the supplies chain loves it (not).
• I always thought Dell posed more of a threat, they do in the US, but not really on the world stage.
• Instead, Lenovo is on a huge growth trajectory! It looks like Lenovo will actually pass HP in either Q3 or Q4 of 2012.

Interesting. Now let's layer tablet shipment figures from IDC to see how they compare:

And I thought Lenovo was growing fast! That's unbelievable growth. But if the iPad, which is 60% of those tablet numbers, is indeed a PC, I would expect to see some sort of decline. I don't really. I can hear Matthew now telling me that it's the same as "buying a secondary PC". Really? That's what netbooks were supposed to be, and their growth numbers looked nothing like this. Instead, these look more like smart phone growth numbers, which are decided NOT PCs.

I firmly believe that tablets and PCs are going to easily co-exist. Instead people are going to give something else up (an iPod Touch perhaps...). The current trends agree.

Now, one of the things I like the Mythbusters is they wouldn't stop here. They would say, "OK. The myth of iPads equalling PCs is BUSTED. But let's assume they are. What would that look like?"

Doing some quick math, I agree with Canalys' conclusion. In Q4 HP shipped 14.7M PC's. That's a lot. In Q3 Apple shipped 2.3M Macs in the US and 11.1M iPads WW (13.4M total). With all the talk of huge iPad estimates for Q4 (at least 20M iPads), it's very feasible that Apple beats HP in Q4 and will continue to do so in 2012. Again, that's IF you count tablets as PCs.

Finally, I took a bigger step back and looked at total PCs vs Tablets:

Clearly tablets are growing like crazy, but they're still a very small piece of the overall pie. It'll still be about 3 years before tablets over-take PCs. (Which is genuinely a long time in the tech space)

Will there ever be a time when tablet's start replacing PCs? Probably. I think there will be something of a tipping point. Once cloud storage and productivity apps mature more, you will then see people ditching their PCs completely for tablets. That isn't happening yet, and it still doesn't change the fundamental premise that iPads are PCs, in the same way that microwaves are not ovens.

** Data sources (Yeah, I know...):
http://www.gartner.com/it/page.jsp?id=1893523
http://www.gartner.com/it/page.jsp?id=1821731
http://www.gartner.com/it/page.jsp?id=1744216
http://www.gartner.com/it/page.jsp?id=1632414
http://www.gartner.com/it/page.jsp?id=1279215
http://www.gartner.com/it/page.jsp?id=1207613
http://www.gartner.com/it/page.jsp?id=1076912
http://www.gartner.com/it/page.jsp?id=939015
http://www.gartner.com/it/page.jsp?id=856712
http://www.gartner.com/it/page.jsp?id=777613
http://www.gartner.com/it/page.jsp?id=724111
http://www.gartner.com/it/page.jsp?id=648619
http://www.gartner.com/it/page.jsp?id=584210
http://www.gartner.com/it/page.jsp?id=1451742
http://www.gartner.com/it/page.jsp?id=1519417
http://www.idc.com/about/viewpressrelease.jsp?containerId=prUS22660011
http://www.gartner.com/it/page.jsp?id=1401136
http://www.gartner.com/it/page.jsp?id=1353330

http://www.idc.com/getdoc.jsp?containerId=prUS23228211
http://www.idc.com/getdoc.jsp?containerId=prUS23034011
http://www.idc.com/getdoc.jsp?containerId=prUS22933011
http://www.idc.com/getdoc.jsp?containerId=prUS22737611

Analyzing Steps - Part Deux

HP just finished sponsoring another shape-up challenge that involves counting steps and exercise minutes. Like normal, I couldn't help myself and kept track on my own. Check it out. The goal (green line) is to hit 10,000 steps a day. The red line is the minimum of 5,000 steps. Yeah... there are more dots below the red line than above the green...

A couple worthy points:

A: Jessi and I spent the day painting the garage of our duplex (finished pictures).

B: Beach Olympics! I was actually sore the following day. :)

Here's an aggregated view by the day of the week. Weekends are good. Thursdays I apparently don't move around much. Last time, Monday and Wednesday were my slow days, this time around they were OK.

Fun stuff. I'll probably take a break from tracking for now and pick it up again later for another round.

You can also check out the results from last time.

Music Library Stats

For the fun of it, I tried copy & pasting my iTunes library into Excel. In wonderful delight, I found that it sort of worked. Well, I couldn't stop there. I created a pivot table and whipped up these views. First, here are some stats:

Timeframe: September 2010 - Yesterday (basically, since I switched to Mac)
Number of Songs in library: 829
Total number of songs played: 9,084
Total play time in library: 2.2 days
Total Computer Storage required: 4.25 GB
Average Times a song is played: 11

Cool. Let's check out a distribution:
(PS. This is my first time using Excel on my mac for something like this. It was a fun experience.)

First, every song has been played at least once, with most number of plays being 56. Second, there's a huge spike at 10. That isn't random - let me explain. iTunes has a feature called "iTunes DJ" where it loads in songs automatically, semi-randomly, playing "popular" songs more. Popularity is determined, as far as I can tell, by the rating you give it (stars) and by how many plays it has. What this means is that if you play a song a lot, it'll get played more and more and more. Furthermore, a rating of 1 star is more "popular" than not being rated at all (0 stars).

So here's what I do: First, I rank every song as 3 stars, and then change it over time. This way 1 star really is the worst.

Second, I created a smart playlist that only contains songs that have less than 10 plays. That means, when a song hits 10 plays, it leaves the playlist. I choose this playlist all the time when using iTunes DJ (which is a majority of the time). As a result, I have a huge spike at 10 because the song exits the playlist and doesn't get played again for a while. Once all my songs are at 10, I'll probably move up to 20 and capture a significant chunk of my songs again.

Next, lets look at genres:

The majority of my library is Christian and soundtrack (which if there were sub-categories, Disney would be the largest). Pop/Rock, Rock & Holiday also make a good showing. I also have one genre named "genre". Clearly a tagging mistake I should fix.

Finally, let's look at some specific artists. The first table is sorted by the number of songs in my library by that artist. The second table is sorted by the number of times I've played a song by that artist.

Well, my mom would be proud to see Kenny Loggins in the #1 slot. There also appears to be a pretty high correlation between the number of songs in my library and the number of times played - only 2 from each list (the *) are not in the other list. Also notable, Hillsong United is a recent addition to my library, so I'm not surprised to see the low play count.

That's it. There's your Monday morning stats. :)

By the way, my #1 played song is "I Will Follow" by Chris Tomlin at 56 times.

Corvallis 2011 Half-Marathon Infographic

Jessi and I participated in the inaugural Corvallis half-marathon this month. I saw the published results and thought something cool could be done with all that data. I've always enjoyed infographics, and wanted to try making one - this seemed like a perfect opportunity. So, here are 13 facts in honor of the 13 mile race.

(You can click the image to see it full-size and/or download it)

For those of you who like to know such things. I housed all the data in Excel (it was originally in a PDF, I'm so thankful for the text-column feature in Excel) and created views using a pivot table. I made the final charts in Excel and copied them into PowerPoint. I added my own comments and pictures - nothing too fancy. I saved the final slide as a PNG and uploaded it to my server. And now you're reading it somewhere in the world. Pretty neat!

Analyzing Steps

HP has been sponsoring a shape-up program for the last 12 weeks. We report our steps and minutes exercised onto a online form. There's lots of cool features, but one missing was a way to analyze your activity. So, I exported my data into Excel (and by "export" I mean typed it in myself) and did a little bit of my own analysis. I thought the findings were cool and wanted to share. All the data is in totals (steps & minutes), but you can still do a relative compare since it's all over the same time frame.

First up are my steps by the day of the week.

Since I have a desk job, I wasn't too surprised that my week-day steps were lower. On Tuesdays I mow the lawn which accounts for the slight bump. My goal was to take a minimum of 5,000 steps a day, but aim for 10,000 steps a day. I hit my 5,000 goal every day but Monday and Wednesday. Saturday was the only day I hit my 10,000 goal.

When I exercise, I take my pedometer off. Most of the time (if not all), these are an official sport I'm playing. Throughout this time I played soccer, volleyball, jiu jitsu and softball.

This data was exactly as I expected. I do jiu jitsu and Wednesday, then split between Monday and Thursday. Soccer and softball happens on the weekend. Since I mow the lawn on Tuesday, I don't feel bad that I'm not doing anything. Friday is really my only free night, and Jessi and I often take that night to go on a walk and talk about our week. Over all, I'm satisfied with my exercise level.

I thought it would be interesting to look at it across the month. July isn't done yet, but enough time has passed that I think it's OK to look at.

Looking at both, I don't know what happened in July. I think it has to do with summer. I've been going on trips, which tends to increase my steps and lower my exercise because I'm out of town.

I haven't fully decided yet, but it would be fun to continue to track my data and see how it changes during the year. I need to find an easy way to do it, or it won't happen.

Finally, here's the raw step data over time.

Do you see the huge spike? That's over 26,000 steps! Jessi and I did that while on a camping trip to Little Crater Lake. Otherwise, you can see I really have two types of days. Either I take 5,000 or less steps, or I take around 10,000 steps. What I really need to focus on is moving those lower numbers up just a little bit more. Perhaps I can put some sort of plan together to add 1,000 steps during the week-day.

Here's what my exercise over time looks like. The data is more sparse, but that's because I don't play a sport every day.

90 minutes is jiu jitsu. 60 was soccer and volleyball. 75 is softball. Some of the higher numbers are days when I had multiple activities. Again, I'm pretty happy about this.

So that's it. Pretty cool findings. I need to work on taking more steps during the week-day, especially Monday and Wednesday. I like my level of exercise and need to just keep it going.