4 ms·

> At the end of the day, making a change in an Excel sheet is easy; understanding formulas is achievable; but learning to code is hard. This is the crux of the

> At the end of the day, making a change in an Excel sheet is easy; understanding formulas is achievable; but learning to code is hard. This is the crux of the matter. I would guess that there are more than an order of magnitude Excel users than Python programmers. Python is great if you already know programming, but expecting domain experts to learn Python in large numbers is going to be a daunting barrier.

indymike 6y ago
I think there's a realization that some computation has become too complex or too risky to put in a spreadsheet. The reasons the author gave for moving to python less about difficulty of coding, but that other problems were bigger than learning to code.
- etothepii 6y ago
  This is exactly the point, do you know of any companies that have had this realisation. If they are based in Bermuda I'd love to work with them.
petepete 6y ago
This is exactly the niche that Resolver One used to fill. It was basically an Excel-like spreadsheet but under the hood everything's Python. Unfortunately it looks like they ceased the product in 2012 due to lack of sales. Perhaps they were too early. https://youtu.be/u6EV2jiKRfc https://youtu.be/u6EV2jiKRfc
- klelatti 6y ago
  Here is sort of a modern variation - upload your Excel workbook to the cloud. I don't have hands on experience or any connection but I think it translates to C# under the hood. https://www.milliman.com/en/products/milliman-mind https://www.milliman.com/en/products/milliman-mind
bash-j 6y ago
This has frustrated me as a python user for the last 7 years, working as the only python user in business environments dominated by Excel. People will say things like, "if you leave, who can support this report you made in python?" Well I say, who can support the bloated 40mb spreadsheet that would take forever to unpick and figure out how to update with new data? No one can, because I've seen people would rather rebuild their own spreadsheet from scratch, than use the files they inherited from the last person. If these tools are necessary to conduct business and they are so worried about being able to support it, why don't they use proper software for that process? A lot of people who make these bloated spreadsheets are people with no education in computing, and don't think about the basics of how to store data that is easy to analyse later. If they are building a weekly report, they build the report and enter the data directly into the report structure, which then makes it almost impossible to analyse later. Next week they just copy the file, rename it and update the data. If you want to analyse that same data over a year, good luck! You can't even count on the data being in the same place over the 52 weeks, since they would have added and removed data points over time. Once I got the process down in a jupyter notebook, handling all the oddities with the data coming from whichever website, CSV file, data warehouse report I need, I can just save it as a .py file and run it as a scheduled task on a virtual computer forever. The data is kept in a format that can be appended to with each update, and can be easily analysed later. The most amazing thing with replacing excel with python is you don't need to manually perform the update process yourself. Which means it doesn't cost anything to run the process more often. Weekly reports can become daily, or even hourly email updates that are only sent when something interesting happens. People can start reacting to things shortly after they happen, rather than having to remember what happened a week or a month ago. The iteration on improving becomes so much faster. People spend more of their time discussing how to fix problems, rather than spending time building problem finders. You can even start to automate the fixing of the problem in python and people don't even have to spend time on that thing at all, ever again.
- kqvamxurcagg 6y ago
  I've used excel and python in lots of business contexts. For most tasks involving domain experts, excel usually wins hands down. An excel spreadsheet is usually easily auditable. The visual presentation and layout lends itself to review by others. You can click and point at values. Python and other programming languages require an environment and tooling that can't be easily supported across the enterprise. It requires source control systems and code review. Programming languages are also too "dynamic". Using excel I can bring in a hard-coded report and link to those values in another tab. In python I'll have to save those to another file or re-query the data source, which may have changed due to new values being retrospectively added. Python is the right tool for lots of analysis tasks, but for most corporate reports it's hard to beat excel. Programmers are also more expensive than corporate analysts. So you would end up replacing teams of low-cost high-retention analysts with high-cost low-retention programmers.

indymike 6y ago

I think there's a realization that some computation has become too complex or too risky to put in a spreadsheet. The reasons the author gave for moving to python less about difficulty of coding, but that other problems were bigger than learning to code.

etothepii 6y ago
This is exactly the point, do you know of any companies that have had this realisation. If they are based in Bermuda I'd love to work with them.

petepete 6y ago

This is exactly the niche that Resolver One used to fill. It was basically an Excel-like spreadsheet but under the hood everything's Python. Unfortunately it looks like they ceased the product in 2012 due to lack of sales. Perhaps they were too early. https://youtu.be/u6EV2jiKRfc https://youtu.be/u6EV2jiKRfc

klelatti 6y ago
Here is sort of a modern variation - upload your Excel workbook to the cloud. I don't have hands on experience or any connection but I think it translates to C# under the hood. https://www.milliman.com/en/products/milliman-mind https://www.milliman.com/en/products/milliman-mind

bash-j 6y ago

This has frustrated me as a python user for the last 7 years, working as the only python user in business environments dominated by Excel. People will say things like, "if you leave, who can support this report you made in python?" Well I say, who can support the bloated 40mb spreadsheet that would take forever to unpick and figure out how to update with new data? No one can, because I've seen people would rather rebuild their own spreadsheet from scratch, than use the files they inherited from the last person. If these tools are necessary to conduct business and they are so worried about being able to support it, why don't they use proper software for that process? A lot of people who make these bloated spreadsheets are people with no education in computing, and don't think about the basics of how to store data that is easy to analyse later. If they are building a weekly report, they build the report and enter the data directly into the report structure, which then makes it almost impossible to analyse later. Next week they just copy the file, rename it and update the data. If you want to analyse that same data over a year, good luck! You can't even count on the data being in the same place over the 52 weeks, since they would have added and removed data points over time. Once I got the process down in a jupyter notebook, handling all the oddities with the data coming from whichever website, CSV file, data warehouse report I need, I can just save it as a .py file and run it as a scheduled task on a virtual computer forever. The data is kept in a format that can be appended to with each update, and can be easily analysed later. The most amazing thing with replacing excel with python is you don't need to manually perform the update process yourself. Which means it doesn't cost anything to run the process more often. Weekly reports can become daily, or even hourly email updates that are only sent when something interesting happens. People can start reacting to things shortly after they happen, rather than having to remember what happened a week or a month ago. The iteration on improving becomes so much faster. People spend more of their time discussing how to fix problems, rather than spending time building problem finders. You can even start to automate the fixing of the problem in python and people don't even have to spend time on that thing at all, ever again.

kqvamxurcagg 6y ago
I've used excel and python in lots of business contexts. For most tasks involving domain experts, excel usually wins hands down. An excel spreadsheet is usually easily auditable. The visual presentation and layout lends itself to review by others. You can click and point at values. Python and other programming languages require an environment and tooling that can't be easily supported across the enterprise. It requires source control systems and code review. Programming languages are also too "dynamic". Using excel I can bring in a hard-coded report and link to those values in another tab. In python I'll have to save those to another file or re-query the data source, which may have changed due to new values being retrospectively added. Python is the right tool for lots of analysis tasks, but for most corporate reports it's hard to beat excel. Programmers are also more expensive than corporate analysts. So you would end up replacing teams of low-cost high-retention analysts with high-cost low-retention programmers.