Month: May 2018

最大槓桿作用的時刻

（從我的一個喜歡與尊敬的作家，賽斯 高汀）

這就是它傾斜之前的那一瞬間，只要稍微多付出一點努力就能發揮重大的作用。

我們時常都在等著這一刻。在參與的那一天得到真正的回報，為我們的努力帶來最大的影響與機械化的優勢。

但是這是一個神話。

我們的最大的槓桿作用是能做出承諾，用每日的持久性，漸進性，積極性，以及表面上不太費力所做出的成果。

當那槓桿作用起效時，它看起來像我們能隨時隨地把握住好的時機。

Metrics Plan, Part 4

In the book, Bare Bones Change Management: What you shouldn’t not do, Bob Lewis explained the seven must-have elements for any change management effort to have a chance of succeeding. Here are my takeaways from one of the topics discussed in the book.

Bob outlined some tactics to consider when working with the cycle time and throughput constraints.

Improved cycle time is about increasing customer satisfaction while improving quality. Lower cycle time brings about lower fixed costs because fewer assets are tied up by the cycles.

The common trade-off for improved cycle time is reduced throughput.

Organizations can consider the following options to improve cycle time.

  • Shorten the cycle step
  • Reduce inter-step buffers or queues
  • Consolidating steps

Improved throughput is about increasing volume with the same unit of time. Higher throughput brings about scalability and lower incremental costs.

The common trade-off for improved throughput is longer cycle time.

Organizations can consider the following options to improve throughput.

  • Shorten the bottleneck step
  • Add parallelism to the bottleneck step
  • Splitting a bottleneck step

Regression Model for Bike Sharing Using R – Take 2

Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

Dataset Used: Bike Sharing Dataset

Dataset ML Model: Regression with numerical attributes

Dataset Reference: https://archive.ics.uci.edu/ml/datasets/Bike+Sharing+Dataset

For available performance benchmarks, please consult: https://www.kaggle.com/contactprad/bike-share-daily-data

INTRODUCTION: Using the data generated by a bike sharing system, this project attempts to predict the daily demand for bike sharing. For this iteration (Take No.2) of the project, we attempt to use the data available, transform as necessary, and apply the Stochastic Gradient Boosting algorithm to examine the modeling effectiveness. Again, the goal of this iteration is to examine various data transformation options and find a sufficiently accurate (low error) combination for future prediction tasks.

This iteration of the project will test the following four modeling scenarios:

Scenario No.1: Remove the attribute “atemp” since it was highly correlated with the attribute “temp.”

Scenario No.2: Perform one-hot-encoding on the variable “season.”

Scenario No.3: Perform one-hot-encoding on the variable “mnth.”

For scenarios 2-3, steps from section No.3 and No.4 will be repeated for each scenario.

CONCLUSION: The baseline performance of the Stochastic Gradient Boosting stands at an RMSE value of 1240 and an R-square value of 0.5943 using the training data. Scenario No.1 did slightly better with an RMSE value of 1233 and an R-square value of 0.5991. As the result, we will leverage scenario No.1 to training the final model and observe how it will do with the validation dataset.

The final Stochastic Gradient Boosting model processed the validation dataset with an RMSE value of 1180 and an R-square value of 0.6320, which was slightly worse than the Take No.1 result of 1177 for RMSE and 0.6329 for R-square. For this iteration of the project, data transformation did not improve the model performance with a noticeable outcome.

The HTML formatted report can be found here on GitHub.

煽動性

（從我的一個喜歡與尊敬的作家，賽斯 高汀）

正確的答案可能不是你能給的最好的建議。

如果你能幫助你的朋友採取行動，那也許會更好。找尋與去做的行為通常總是會比獲得好的建議來更有用。

煽動行動往往會比提供洞察力來的更好。要更好地繼續前進，找出解決問題的辦法，而不是停下來因為你相信你知道最正確的答案。

Creating Scarcity, Part 1

In the podcast series, Seth Godin’s Startup School, Seth Godin gave a guided tour to a group of highly-motivated early-stage entrepreneurs on some of the questions they will have to dig deep and ask themselves while they build up their business. Here are my takeaways from various topics discussed in the podcast episodes.

  • When people choose to do business with someone or an organization, they are asking whether they are willing to hire the product or service to solve a problem they have. Often our problem is more of emotional by nature. We want to impress our boss. We want to impress or do good for our family. We want to connect with someone or something because we are lonely. We want something that can reassure us because we are afraid.
  • Differences between marketing and manipulation are… Manipulation is where we get people to buy something, and they regret after they bought it. Marketing is where we get people to buy something, and they are glad they did afterward. The difference is the “story” we tell our prospective customers, and the story must be ethical and resonate with the people who need to hear it.
  • If what you do is catching on, what is preventing someone else provide the same product/service at half of the price? The solutions are the lock-in and network effects.
  • The network effects form when something you built works better when more people are using Fax, email, are social media examples of something works much better with the network effect. When designing our products and services, we should consider how to build features that can encourage the network effect.
  • When thinking of pricing and value creation, it is useful to think of a leaky funnel with the paying customers coming out at the bottom. There are costs involved every step of the way from acquiring prospects and to turning them into paying customer.
  • “What problem is my customers waking up every morning with?” is a useful question to ask. It all comes down to… Do people know you? Do they trust you? Will they listen to your story? Does their story resonate enough with yours, so they are willing to give you a try?
  • Once we have accumulated a viable customer base, our job is not so much about finding more customers. Our job is to find more useful and interesting products for our customers. We want to leverage the connection economy to have our customers talk about us and market our products/services for us. Furthermore, we want to leverage the connection and move into the position of being the arbiter of the community standards, which is becoming a world of one.

Networks, Lock-in, and Pathways

In his podcast, Akimbo, Seth Godin teaches us how to adopt a posture of possibility, change the culture, and choose to make a difference. Here are my takeaways from the episode.

  • Gillette figured out that selling one razor is a tough business, but selling a lifetime of razors is a great business.
  • Network, Lock-in, and Pathway effects are important to understand for two reasons. First, if we are not careful with them, we might fall as a victim to those effects. Two, those effects are powerful ways to dance with culture, especially when we are trying to affect changes on the culture.
  • Metcalfe’s law states the effect of a telecommunications network is proportional to the square of the number of connected users of the system (n2). Another word, more connections create value. That is the network effect.
  • Lock-in occurs when the cost of switching from one solution to another feels too much. Marketers and businesses work hard to create those effects when we use their products and services.
  • The work required to create the lock-in effect and path dependency is immense and slows you down at first. Once the structure is in, it is much easier for others to plug in and become interoperable with the system as it currently exists.
  • On the other hand, for freelancers and entrepreneurs who want to affect changes, the easy path of being plug-and-play and interoperable is not the way to change the culture, to create value, and to make the impact.
  • The lock-in effect can also make it difficult to change minds or reconsider past choices by creating a cognitive dissonance thinking.
  • How to launch something new when there is already something in place that is dominant? Find something outside of where you are that is a fundamental shift, which encourages new choices to be made. At that moment, all the lock-in and sunk costs seem less important because the shift creates a new network effect with momentum. Find those moments of the shift.
  • During those moments of the shift, Seth suggested four questions we can ask ourselves.
    1. What are the sunk costs that are getting in the way of you making the new decision?
    2. When you commit to a solution, do you take the approach of open systems (which lessen the switch costs later) or do you take close system (which has the lower upfront costs but with lock-in)?
    3. What is the sea of change coming to your industry that is going eliminate many of the lock-in, path-dependencies, and sunk costs? How will that change allow you to have a fresh start?
    4. Are you willing to invest in creating a network effect, in building something better that will bring more value when more people have access to your solution? After you can create a network effect, a cultural shift becomes possible.
  • The things we love and trust all have a network effect. “People like us do/use thing like is” Culture is the ultimate network effect. It is on us to speak up about the network effect, the should’s and should not’s, and it is up to us to use the network effect wisely.

Regression Model for Bike Sharing Using R – Take 1

Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

Dataset Used: Bike Sharing Dataset

Dataset ML Model: Regression with numerical attributes

Dataset Reference: https://archive.ics.uci.edu/ml/datasets/Bike+Sharing+Dataset

For available performance benchmarks, please consult: https://www.kaggle.com/contactprad/bike-share-daily-data

INTRODUCTION: Using the data generated by a bike sharing system, this project attempts to predict the daily demand for bike sharing. For this iteration of the project, we attempt to use the data available for discovering a suitable machine learning algorithm that future predictions can use. We have kept the data transformation activities to a minimum and drop the several attributes that do not make sense to keep or simply will not help in training the model. Again, the goal of this iteration is to find a sufficiently accurate (low error) algorithm for the future prediction tasks.

CONCLUSION: The baseline performance of predicting the target variable achieved an average RMSE value of 1322. Three algorithms (Bagged CART, Random Forest, and Stochastic Gradient Boosting) achieved the lower RMSE and higher R-square values during the initial modeling round. After a series of tuning trials with these three algorithms, Stochastic Gradient Boosting produced the lowest RMSE value of 1213 and the highest R-square value at 0.6093 using the training data.

Stochastic Gradient Boosting also processed the validation dataset with an RMSE value of 1177 and an R-square value of 0.6329, which was better than the average training result. For this project, the Stochastic Gradient Boosting ensemble algorithm yielded consistently top-notch training and validation results, which warrant the additional processing required by the algorithm.

The HTML formatted report can be found here on GitHub.

創業不是一份工作

（從我的一個喜歡與尊敬的作家，賽斯 高汀）

你不必去申請。你不會得到工資。也沒有人會來挑選你。

當你大肆吹噓的你籌集了多少錢，或者你的估值是多少，那是一種辦公的思維。

創業是做出一個將解決的方案給一個有問題需要解決的人來做的交易機會。

能解決更多問題，能解決更大的問題，能更廣泛地解決問題，那你就是企業家。

將一項工作實現工業化，使之與規則，老闆和流程相適應是很有誘惑力的。但這不是創業的核心。

創業是需要以你感到自豪的方式來解決問題。