文章目錄
6 結合團隊動力學的模型拓展
在對於哈士奇球隊的研究中，我們瞭解到一些影響成功團隊合作的因素，例如傳球網絡、個人能力、教練等等。這些因素可以和團隊動力學進行聯繫，用來分析爲什麼這些足球方面的因素對球隊的表現有貢獻。以及我們可以探索考慮什麼樣的因素可以補充到我們解釋優秀團體，並推廣到社會上各種各樣的團體。
6.1 團體動力學與足球隊
團體動力學主要包括五個方面的內容：團體內聚力，羣體壓力和羣體標準，個人動機和羣體目標，領導與羣體性能以及羣體的結構性。
Group dynamics mainly includes 5 aspects: group cohesiveness, group pressure and group standards, individual motivation and group goals, leadership and group performance and group structure.
6.1.1 羣體內聚力 group cohesiveness
羣體內聚力可以指球隊想贏的信念，羣體壓力是來自外界的壓力。提高羣體內聚力是提高球隊贏球信念的良性循環。責任性行爲體現在每個球員各司其職。成員之間的相互影響表現在球員間的相互鼓勵，相互進步。
Group cohesion can be regarded as the belief that the team wants to win. Group pressure comes from the outside. Improving group cohesion is a virtuous circle to inspire the team’s winning belief. Responsible behavior is reflected in each player’s performance. The mutual influence among the members is reflected in the mutual encouragement and progress among the players.
Huskies在比賽中承擔着對手的壓力，同時也充滿着求勝的鬥志。在Huskies球隊的數據分析中，每個人場均acceleration、跑動距離和熱點圖的有效面積等數據都可以反映每個人在球場上的態度和努力。在落後、平局、領先的情況下，團隊承受着不同的外在壓力，球員反映態度的數據也會波動，若能在平局和落後的情況下依然有着不遜領先時候的positive數據，那我們可以認爲羣體內聚力強大。
The Huskies is under the pressure of their opponents in the competition, but also full of victorious spirit to win. In data analysis of Huskies, each players’ average acceleration, running distance and effective area of heatmap can reflect the particular player’s attitude and effort on the court. In the
situation of backwardness, draw and lead, the team bears varying degrees of external pressure, as the result, the data of players’ attitude will fluctuate. If there is still positive data in the situation of draw and backwardness, we can consider that the group cohesion is really strong.
6.1.2 羣體標準和羣體壓力 Group standard and Group pressure
羣體標準可以幫助球隊和球員更有壓迫感，相互競爭是提高個人能力，由此觀之，適當的羣體壓力是有必要的，有競爭纔會有發展。
Group standards can help teams and players to feel more oppressive and to compete with each other, which helps improve personal ability. From this point of view, appropriate group pressure is necessary, and only when there is competition can there be development.
在Huskies球隊中存在出場時間的差異，核心球員和邊緣球員的出場時間差距巨大，導致了邊緣球員在球隊中會受到核心球員技術、地位上的壓力，但這也激勵了他們努力獲得出場時間來證明自己。可以分析每個球員每一場比賽評價的趨勢，尤其是不能獲得穩定出場時間的球員們，他們的評價若能在有限的出場中得到一定提升，則可以認爲對內競爭帶來的壓力使他們進步。
For Huskies, there are differences in minutes. There is a huge gap between the core players and the edge players, which leads to the edge players in the team will be under the pressure of the core players’ ability and status, but it also encourages them to strive to get playing time to prove themselves. We can analyze the evaluation trend of every game of players, especially the players
who can’t get stable playing time. If their evaluation can be improved in limited playing chances, we can think that the pressure brought by internal competition makes them progress in a way.
6.1.3 個人動機和羣體目標 Individual motivation and group goals
羣體目標影響羣體行爲，當球隊的目標和球員的目標達成一致時，球員會表現出最爲強烈的求勝動機，努力爲進球而拼搏。
Group goals affect group behavior. When team goals and players’ goals are consistent, players will show the strongest motivation to win and strive for goals.
球隊每個位置的球員有不同的職責，而勝利則是所有球員都較爲完美地完成了自己的任務。因此每個人在勝利的目標上是相同的，而且每個人對自己的職責存在一定的期望，當所有人同時爲着自己任務完成的期望而努力，羣體目標便達成一致。
Players in every position of the team have different responsibilities, and the victory is that all players have completed their tasks perfectly. Therefore, everyone is the same in the goal of victory, and everyone has certain positive expectations for their own tasks. When everyone works hard for the expectation of completing their own tasks at the same time, the group goals will reach an agreement.
6.1.4 領導與羣體性能 Leadership and group performance
在球隊中，領導者有兩種，可以是爲球員提供訓練，賽前指導和賽後總結的教練，也可以是在賽場上鼓舞士氣，樹立榜樣和調度指揮的隊長。領導者的能力會影響球隊的發揮和球隊的進步。此外，教練和隊長對球員的關照可以很好地鼓提高隊內活力。
There are two kinds of leaders in the team, they are the coaches who provide training, pre-match guidance and post-match summary for the games, or the leaders who inspire morale, set an example and dispatch the command on the field. The ability of leaders will affect the team’s performance and progress. In addition, the coach and the captain’s help to the players can improve the team’s vitality.
6.1.5 羣體的結構性 Group structure
當一個球隊在球員關係穩定時，擁有了一定的球隊結構。穩定的結構有利於培養球員間默契，從而有更大的概率贏得球賽。可以認爲團隊不同高低的地位的人數服從上三角形型分佈，而若地位極差太大或太小，都不利於團隊的穩定性。此外，若球隊某一位置競爭極大，而其他位置競爭極小，則會產生結構不平衡的問題，引發結構變化。
When a team has a stable relationship between players, it has a team structure. Stable structure is conducive to the cultivation of tacit understanding among players, so as to have a greater probability of winning the game. It can be convinced that the number of people in different positions of a team obeys the upper triangle distribution, and if the position range is too large or too small, it is not conducive to the stability of the team. In addition, if the competition in one position of the line-up is too great and the competition in other positions is very small, there will be structural imbalance and structural change.
6.2 成功團隊合作其他影響因素 Other influence factor of successful teamwork
團隊之間還需要有密切、暢通的信息交流網。國際化球隊需要確保球員之間的語言交流，能夠在球場內外溝通順暢；此外還要多進行聯誼活動，提升隊內融洽氛圍。這樣才能在比賽時做到有效的交流和配合。
There is also a need for close and unblocked information exchange network between teams. International teams need to ensure that the language communication between players can be smooth on and off the court; in addition, they need to carry out more friendship activities to improve the harmonious atmosphere within the team. In this way, we can communicate and cooperate
effectively in the competition.
領導和每一個人都應該尊重他人與自己間存在的不同，在保持羣體目標和內聚力的情況下，通過接受人們之間的差異來改善團隊氛圍。
Leaders and everyone should respect the differences between others and themselves, and improve the team atmosphere by accepting the differences between people while maintaining group goals and cohesion.
管理層要能清晰地認識團隊內部的情況，並且有能力爲現狀變化做調整。
Administrators should have a clear understanding of the situation and be able to make adjustments to the situations.
7 評價
7.1 優勢 Strength
- 傳球評價指標PEI的設計和傳球網絡模型PNM建立緊密相關。PEI綜合考慮了每一次傳球的多個方面，量化了傳球的好壞程度，能縮小與實際情況誤差和方差；傳球網絡模型以圖論作爲基礎，直觀描述配合默契程度，有助於多元組合的搜索，並且可視化的效果能突出默契組合。
The establishment of PNM is closely related to the design of PEI. PEI
comprehensively considers many aspects of each pass, quantifies the quality
of the pass, and can reduce the error and variance with the actual
situation. And the network model of pass, based on graph theory, intuitively
describes the degree of cooperation, which is conducive to the search of
multiple combinations, and the visual effect can highlight the familiar
combination.
- 熱力圖生成模型對於離散型數據的近似連續化有很強的兼容性，能夠應對座標數據過少或稀疏的情況，基於可視化的數據分析球員站位與實際情況相符；對於events數據的清洗有效規避了數據異常和缺失所帶來的影響，建立隨機森林分類器模型在比賽樣本數量僅爲38的情況下不容易過擬合或偏差過大，參數調優後最高達到80%的準確率足以對比賽大致結果進行有效預測，what
means該模型可以基於近期數據爲未來比賽做出預測並且給教練的訓練和line-up作爲參考。
Heatmap generation model has strong compatibility for the approximate
continuity of discrete data, and can cope with the situation of too little
or sparse coordinate data. Based on the visual data analysis, player
position is consistent with the actual situation. In addition, for the
cleaning of events data, the impact of data abnormality and missing is
effectively avoided. Even though the number of samples is only Under 38, it
is not easy to overfit or deviate too much. The highest accuracy of 80%
after parameter optimization is enough to effectively predict the general
results of the game, what means this model can make predictions for the
future game, that reflects the current ability of the team, based on the
recent data and give training and line up to coaches as reference.
- 將教練的靜態結構策略轉化爲最優排列組合問題；大規模數據支持了評價指標的維度增加，降低了評價球員位置能力值偏差的期望；模擬退火算法在算力有限的情況下，依據實際經驗設置啓發策略，運用無法精準的個人能力評價指標來尋得11人球員排列組合的局部最優解，在準確度期望一定的情況下，可以在閾值範圍內接受它爲全局最優解。
The static structure strategy, which should be developed by the coach, is
transformed into the optimal arrangement and combination problem of 11
elements. Large-scale data supports the increase of the dimension of the
evaluation index and reduces the expectation of the deviation of the
particular position ability value of each player. Under the condition of
limited computing power, the simulated annealing algorithm is properly used,
and we manually set the starting strategy according to the actual experience
and uses the imprecise individual ability evaluation index to find the
arrangement and combination of 11 players. The partial optimal solution can
be accepted as the global optimal solution within the threshold range when
the accuracy expectation is certain.
- 上述的模型能夠很容易地對應到group
dynamics中的理論關鍵點，基於已有影響因素的額外方面也具有很大實際意義。
The models can easily correspond to the theoretical key points in group
dynamics, and the additional aspects based on the existing influencing
factors are also of great practical significance.
7.2 缺陷 Weakness
- 在模型中超參數較多，模型的參數調優有着極大的挑戰。
There are many hyper parameters in the model, so the parameter optimization
of the model has a great challenge.
- 每個時刻有數據的球員較少，對球員無球能力的考查無法進行，且難以從整體的站位和陣型評價進攻或防守。
There are fewer players with data at each time, so it is impossible to test
the players’ ability without the ball, and it is difficult to evaluate the
attack or defense from the aspect of the overall position and formation.
- 輸入隨機森林分類器模型的比賽樣本過少，訓練結果波動較大。
There are too few samples to input the RFC model, so the training results
fluctuate greatly.
- 模擬退火算法得出的最佳陣容僅能確保爲局部最優解，不能保證爲全局最優解。
The optimal lineup obtained by the simulated annealing algorithm can only be
guaranteed to be a local optimal solution, not a global optimal solution.
全文：