Algorithmic Life (11): Looking at the "2/8 Time Management Method" from the "Gradient Boosting Tree (GBDT)"

Combined with GBDT (Gradient Boosting Tree) and the 2/8 rule, how can we better practice time management?

Gradient Boosting Decision Trees (GBDT) is an efficient machine learning algorithm that improves the accuracy of predictions by iteratively building multiple decision trees and synthesizing their prediction results. The core idea of GBDT is that each tree tries to correct the predicted residuals left by the previous tree, and in the construction process, each feature contributes to the improvement of model performance by splitting nodes, so that the importance of each feature can be evaluated as follows:

1. Train a model: Use the training dataset to train a gradient boosting tree model, and multiple decision trees will be generated during the model construction process. GBM or other models that support feature importance evaluation are used for training, and the fitting results of the model to the training dataset are obtained.
2. Calculate feature importance: Extract the importance score of each feature from the trained model. The importance of a feature is measured by calculating its contribution to the reduction of the loss function when it acts as a split node in all trees. Each time each feature is selected as a split node, it leads to an increase in the model's fit to the training data, i.e., a decrease in the loss function. The greater the reduction of the loss function, the greater the effect of this feature on the performance of the model.
3. Sorting features: After the construction of multiple trees, the importance score of each feature in the whole model is summarized, that is, the sum of the loss function reduction of each feature as a split node, and the total importance score of the feature in the whole GBDT model is obtained by accumulating. Features are sorted in descending order based on feature importance scores.
4. Feature selection: Visualize the importance of features, such as histograms and bar charts, to visualize which features contribute the most to the model's prediction performance. According to the idea of the Pareto rule, that is, a small number of features contribute most of the model performance, then the top-ranked features can be preferentially selected for model training based on these importance scores (the proportion can be adjusted according to the specific situation).
5. Validation and optimization: Retrain the model with a subset of selected features and validate model performance through cross-validation or other evaluation methods, and fine-tune the number of feature selections by incrementally increasing or decreasing features to achieve optimal model performance.

In addition, other feature selection methods, such as linear models based on L1 regularization (such as Lasso), also exhibit a phenomenon similar to Pareto's rule in the process of compressing sparse solutions, that is, a large number of features are clipped, leaving only a few features that are crucial for model prediction.

From the above introduction, we can see that a very important idea in GBDT (Gradient Boosting Tree) is to "extract important features", and focus on "features that have a significant impact on the output of results", rather than "taking care of all features", which can not only save the energy of training the model, but also effectively reduce the complexity of the model and improve the interpretability of the model. This idea follows a business rule: the 2/8 rule.

The rule of 2/8, also known as the Pareto rule, the rule of 80/20, or the rule of the critical minority, is that in many cases, 80% of the results are due to 20% of the causes, or that 20% of the effort produces 80% of the output. This law is common in many fields such as society, economy, and management. However, it is not only the idea that the 2/8 rule can be applied to business thinking, but also the personal thinking, such as the "2/8 time management method".

To apply the 2/8 rule in time management, the following steps can be followed:

1. Make a list of tasks: First, you need to make a list of all your current tasks, making sure that they are comprehensive and that nothing is missing. Also, make a list of the key goals that the individual needs to accomplish the most at the moment.
2. Assess the importance of tasks: Each task is evaluated, ranked in order of its importance to accomplishing key objectives. This step is critical because it determines which tasks are really important and need to be prioritized.

3. Identify key tasks: Identify the critical 20% of the tasks in the to-do list, which tend to be the tasks that have the most impact on the key objectives. At this time, it is necessary to learn a certain "disconnection", that is, not to intuitively think about "which tasks have the greatest impact on the goal", it is best to have quantifiable data to help you identify.

   For example, the current key goal is to "quickly improve your English speaking ability in the short term", and the tasks that can be classified are: memorizing words, taking online classes, reading English news, etc., then you should analyze the best way to improve your spoken English in the short term according to your current state, vocabulary base, specific content of online classes, etc. You can try and record the impact of each item on "speaking improvement", and then decide "which one you want to focus on the fastest way to improve your spoken English" according to your current state and goals.

   The author recommends setting only one main task at a time, and then choosing the next most important task when this task is completed (depending on the nature of the matter, you can adjust it yourself, mainly to "focus" as much as possible, which helps to improve efficiency!)

4. Allocate time blocks: Divide the daily working time into several time blocks, put these 20% of the key tasks in the first place (the proportion can be adjusted according to the situation), prioritize the allocation of time and energy to deal with them, and ensure that these tasks are completed with high quality (try to use a large block of time to do important tasks, and a long time is more conducive to entering the "flow", and "entering the flow" is more conducive to the improvement of completion efficiency). If there are multiple critical tasks to be dealt with, the priority can be set based on urgency and importance. For the remaining 80% of tasks (the proportion can be adjusted), allocate time reasonably, try to avoid wasting too much time on trivial things, and prioritize ensuring that key tasks are prioritized.
5. Eliminate distractions: When dealing with critical tasks, distractions should be avoided as much as possible. This can be done by silencing your phone, turning off social media, and more to create a focused work environment.

Let's use the role of a programmer as an example of how to apply the "2/8 Time Management Method":

1. Mission-critical analysis and identification: Applying the 2/8 rule means identifying the 20% of tasks that have the greatest impact on the overall progress of the current project and the quality of the product when faced with a wide range of programming tasks, bug fixes, code reviews, technical documentation, etc.

   For example, the main goal at present is to launch a new product as soon as possible, so that users can use the idea of verifying the product, then we can discuss together "what are the key steps for users to use as soon as possible" and "what must be done to complete this key step", that is, at this time, we should discuss "save energy", in line with the principle of "picking", carefully discuss the business scenario, try to "pick and use energy", and let the team's "energy" be spent on the most critical steps.

   Of course, the identification of this key task may mainly depend on the product manager, but in terms of specific technical implementation, if you can also "pick and choose" more, and understand more about the specific business scenarios and the reasons behind them, it will be helpful to find out where you can "save energy" in the requirements discussion stage or code implementation stage.

   Product design needs to be verified by the market and users, and needs to be quickly trial-and-error, and every unnecessary task will consume the energy of the R&D team, so reducing the cost of trial and error as much as possible, identifying key tasks and the implementation of key tasks, is the key to freeing the R&D team from fatigue development. This requires the efforts of the entire R&D team and more thinking from their respective professional perspectives to avoid this.

2. Time allocation: Allocate the most energetic 80% of the day (the proportion can be adjusted, just to illustrate the main time and energy) to the most important 20% of the tasks, and ensure that these key tasks can be completed efficiently and accurately. At the same time, we should also pay attention to the links that frequently cause problems in the work, and optimize and improve them in a targeted manner, which can greatly improve the overall work efficiency.
3. Resource utilization: Identify, refine, and improve the tools, resources, and skills that are most commonly used in the job, as they may lead to the majority of the work. Nowadays, there are many AIGC products, each of which has its own advantages, and you can try more products of each company according to your current task needs to see how to use them, so that you can be freed from tedious work or things that you are not good at but AI is good at, so that you can focus more on the processing of key tasks.

Speaking of the 2/8 rule, there is a recent Yang Zi drama "Chenghuan Ji" is on the air, one of which is in the apartment managed by Yang Zi, there is a family that is difficult to serve the tenant who often asks Yang Zi to do some trivial things, and the reason why Yang Zi is willing to do it is also because only by serving the guest well, can the tenant improve the service quality of the apartment, and then pay the property fee. At this time, the male protagonist reminded Yang Zi "to pay attention to the things that this resident cares about the most, and don't let yourself be tossed all day long by things like walking dogs and buying milk tea."

All the principles in life are the same, whether it is algorithms, work, getting along with people, etc., you can use the "2/8 rule" to think.

There is a detailed description of how the Buddha's disciples cultivated the "six roots" to break the "six dusts", and the "eyes, ears, nose, tongue, body, and mind" are difficult to cultivate, and the roots are entangled, affecting each other, and it is difficult to start. According to their own "practice", the disciples of the Buddha each chose "one" suitable for themselves to cultivate, and then "one is passed, and all six are passed".

The author thinks that this is also very similar to the "2/8 rule", people's life and energy are limited, everything "grasp the big and let go of the small", the time can be "more", the energy is more focused, and the problem can be better solved.

This article was originally published by @养心进行时 on Everyone is a Product Manager and is not allowed to be reproduced without permission

Image from Unsplash, based on the CC0 license

The views in this article only represent the author's own, everyone is a product manager, and the platform only provides information storage space services.